Beruflich Dokumente
Kultur Dokumente
WITHIN THE
BOOK ONLY
<
C0>
-z
60351 >m
OSMANIA UNIVERSITY LIBRARY
Mo. Accession No.
$f)
NEW YORK
THE MACMILLAN COMPANY
1949
COPYRIGHT, 1937,
BY THE MACMILLAN COMPANY.
each takes such a fragmentary view of the whole subject, that his
work is ineffectual.
The world in which the boy and girl have lived is the true
laboratory of elementary mechanics. The tennis ball, the golf
ball, the shell on the river the automobile
; good old Model T,
in its day, and the home-made autos and motor boats which
vi PREFACE
"Fortune favors the prepared mind" that may well be a guide for
us now and in the future. What can be done, and what we havo
attempted in the present work, is to unite a broad and deep knowv
PREFACE vii
spirit scientific.
The book is adapted to the needs of a first course in Mechanics,
given for sophomores, and culminating in a thorough study of the
dynamics of a rigid body in two dimensions. This course may be
followed by a half-course or a full course which begins with the
kinematics and kinetics of a rigid body in three dimensions and
proceeds to Lagrange's Equations and the variational principles.
So important are Hamilton's Equations and their solution by
means of Jacobi's Equation, that this subject has also been in-
cluded. It appears that there is a special need for treating this
rudiments, but hard reading for the beginner through poor pre-
sentation and lacunae in the theory Appell, Mecanique rationelle
; y
CHAPTER II
CHAPTER III
MOTION OF A PARTICLE
1. Motion
Rectilinear 49
2. Newton's Laws of Motion 50
3. Absolute Units of Force 55
4. Elastic Strings 58
ix
x CONTENTS
PA OB
5. A Problem of Motion 60
6. Continuation; the Time 63
7. Simple Harmonic Motion 64
8. Motion under the Attraction of Gravitation 69
9. Work Done by a Variable Force 72
10. Kinetic Energy and Work 75
11. Change of Units in Physics 76
12. The Check of Dimensions 79
13. Motion a Resisting Medium
in 81
14. Graph of the Resistance 84
;
15. Motion in a Plane and in Space 86
16. Vector Acceleration 90
17. Newton's Second Law 92
18. Motion of a Projectile 93
19. Constrained Motion 95
20. Simple Pendulum Motion 97
21. Motion on a Smooth Curve 99
22. Centrifugal Force 101
23. The Centrifugal Oil Cup 105
24. The Centrifugal Field of Force 106
25. Central Force 108
26. The Two Body Problem 114
27. The Inverse Problem to Determine the Force 114
28. Kepler's Laws 115
29. On the Notion of Mass 118
CHAPTER IV
PAGE
14. Billiard Ball, Struck Full 143
15. Continuation. The Subsequent Motion ........ 145
16. Further Examples 146
Exercises on Chapter IV 151
CHAPTER V
KINEMATICS IN TWO DIMENSIONS
'
1. The Rolling Wheel 154
2. The Instantaneous Centre 155
3. Rotation about the Instantaneous Centre 157
4. The Centrodes 159
5. Continuation. Proof of the Fundamental Theorem .... 162
6. The Dancing Tea Cup 165
7. The Kinetic Energy of a Rigid System 166
8. Motion of Space with One Point Fixed 168
9. Vector Angular Velocity 170
10. Moving Axes. Proof of the Theorem of 8 172
11. Space Centrode and Body Centrode 174
12. Motion of Space. General Case 175
13. The Ruled Surfaces 176
14. Relative Velocities 177
15. Proof of the Theorem of 12 179
16. Lissajou's Curves 182
17. Continuation. The General Case. The Commensurable Case.
Periodicity 186
Professor Sabine's Tracings of Lissajou's Curves
between pages 190-191
CHAPTER VI
ROTATION
1. Moments of Inertia 191
2. Principal Axes of a Central Quadric 194
3. Continuation. Determination of the Axes 196
4. Moment of Momentum. Moment of a Localized Vector . . 197
5. The Fundamental Theorem of Moments 199
6. Vector Form for the Motion of the Centre of Mass .... 201
7. The Invariable Line and Plane 201
8. Transformation of 202
9. Moments about the Centre of Mass 204
10. Moments about an Arbitrary Point 205
xii CONTENTS
PAGE
11. Moments about the Instantaneous Centre 207
12. Evaluation of a for a Rigid System ;
One Point Fixed ... 208
13. Euler's Dynamical Equations 210
14. Motion about a Fixed Point 212
15. Euler's Geometrical Equations 214
16. Continuation. The Direction Cosines of the Moving Axes . . 216
17. The Gyroscope 217
18. The Top 220
19. Continuation. Discussion of the Motion 222
20. IntrinsicTreatment of the Gyroscope 225
21. The Relations Connecting v, F, and K 229
22. Discussion of the Intrinsic Equations 231
23. Billiard Ball 237
24. Cartwheels 241
25. R&ume* 245
CHAPTER VII
CHAPTER VIII
IMPACT
1. Impact of Particles 270
2. Continuation. Oblique Impact 274
3. Rigid Bodies 277
4. Proof of the Theorem 279
5. Tennis Ball, Returned with a Lawford 282
CHAPTER IX
RELATIVE MOTION AND MOVING AXES
1. Relative Velocity 285
2. Linear Velocity in Terms of Angular Velocity 285
CONTENTS xiii
PAGE
3. Acceleration 287
4. The Dynamical Equations 290
5. The Centrifugal Field 291
6. Foucault Pendulum 292
CHAPTER X
LAGRANGE'S EQUATIONS AND VIRTUAL VELOCITIES
1. The Problem 297
2. Lagrange's Equations in the Simplest Case 299
3. Continuation. Particle on a Fixed or Moving Surface . . . 303
4. The Spherical Pendulum 306
5. Geodesies 308
6. Lemma 310
7. Lagrange's Equations in the General Case 312
8. Discussion of the Equations. Holonomic and Non-Holonomic
Systems 313
9. Continuation. The Forces 315
10. Conclusion. Lagrange's Multipliers 316
11. Virtual Velocities and Virtual Work 318
12. Computation of Qr 320
13. Virtual Velocities, an Aid in the Choice of the TTT 321
14. On the Number m
of the Q T 322
15. Forces of Constraint 325
16. Euler's Equations, Deduced from Lagrange's Equations . . . 325
17. Solution of Lagrange's Equations 326
18. Equilibrium 330
19. Small Oscillations 333
Exercises on Chapter X 336
CHAPTER XI
HAMILTON'S CANONICAL EQUATIONS
1. The Problem 338
2. A General Theorem 339
3. Proof of Hamilton's Equations 342
CHAPTER XII
D'ALEMBERT'S PRINCIPLE
1. The Problem 345
2. Lagrange's Equations for a System of Particles, Deduced from
d'Alembert's Principle 348
3. The Six Equations for a System of Particles, Deduced from
d'Alembert's Principle 349
riv CONTENTS
PAGE
4. Lagrange's Equations in the General Case, and d'Alembert's
Principle 350
5. Application : Euler's Dynamical Equations 352
6. Examples 353
CHAPTER XIII
CHAPTER XIV
CONTACT TRANSFORMATIONS
1. Purpose of the Chapter 389
2. Integral Invariants 392
3. Consequences of the Theorem 395
4. Transformation of Hamilton's Equations by Contact Trans-
formations 400
5. Particular Contact Transformations 403
6. Theft-Relations 407
CHAPTER XV
SOLUTION OF HAMILTON'S EQUATIONS
1. The Problem and Its Treatment 410
2. Reduction to the Equilibrium Problem 413
3. Example. Simple Harmonic Motion 415
4. H, Independent of t. Reduction to the Form, H' = Pi . . . 420
5. Examples. Projectile in vacuo 424
CONTENTS xv
PAGE
6. Comparison of the Two Methods 429
7. Cyclic Coordinates 430
8. Continuation. The General Case 433
9. Examples. The Two-Body Problem 434
10. Continuation. The Top 438
11. Perturbations. Variation of Constants 440
12. Continuation. A Second Method 444
APPENDIX
A. Vector Analysis 447
z =
B. The Differential Equation (du/dt)
:
f(u) 456
C. Characteristics of Jacobi's Equation 466
D. The General Problem of Rational Mechanics 476
INDEX . . . . 491
MECHANICS
CHAPTER I
STATICS OF A PARTICLE
1. Parallelogram of Forces. By a force is meant a push or a
pull. A stretched elastic band exerts a force. A spiral spring,
likethose used in the upholstered seats of automobiles, when
compressed by a load, exerts a force. The earth exerts a force
of attraction on a falling rain drop.
The effect of a force acting at a given point, 0, depends not
merely on the magnitude, or intensity, of the force, but also on
the direction in which it acts. Lay off a right line from
O in the direction of the force, and mpke the length of ^ ^
the line proportional to the intensity of the force for ;
pJG 1
FBlTKf,
FIG. 4 FIG. 5
EXERCISES
1. Two forces of 5 Ibs. and 12 Ibs. make a right angle with
each other. Show that the resultant force is 13 Ibs. and that it
makes an angle of 22 37' with the larger force.
what angle must they make with each other and with the resultant?
5. A force of 100 Ibs. acts north. Resolve it into an easterly
and a north-westerly component.
6. A force of 50 Ibs. acts east north-east. Resolve it into an
easterly and a northerly component. Ans. 46.20 Ibs. ;
19.14 Ibs.
7. A force of 12 Ibs. acts in a given direction. Resolve it into
two forces that make angles of 30 and 40 with its line of action.
Here, a = P, b = Q, c = R,
C = 180 - w
and hence c
(2) 72
2 = P2 +Q + 2
2PQ cos co. FIG.
R = 7 Ibs.
To complete the solution and find the remaining angles we
can use the Law of Sines :
4 MECHANICS
~ I,
(3)
sin A sin B sin C
Thus
1
(4)
sm ^> sm
There is no difficulty here about the sign when the adjacent
angle is to)
= sin co.
used, since sin (180
In the numerical example above, Equation (4) becomes :
8 7
sin <p |\/3
Thus
sin <p
= 4-V3, cos <p
= | , <t>
= 81 47'.
The third angle is found from the fact that the sum of the
A +B+C = 180.
EXERCISES
Give both a graphical and an analytical solution each time.
1. Forces of 2 Ibs. and 3 Ibs. act at right angles to each other.
* Observe that in this case it is easier to determine the angle from its tangent.
Square roots should be computed from a Table of Square Roots. Huntington's
Four-Place Tables are convenient, and are adequate for the ordinary cases that
arise in practice. But cases not infrequently arise in which more elaborate tables
are needed, and Barlow's will be found useful.
STATICS OF A PARTICLE 5
C = A + B.
It is obvious that
B +A = A + B.
Any number of vectors can be added by applying the definition
(A + B) + C = A + (B + C).
A +B = 0.
We write, furthermore,*
B =- A.
P Q _ E
sin p sin q sin e
"
FIG. 8 p +q+e= 180.
* It not necessary for the present to go further into vector analysis than the
is
above definitions imply. Later, the two forms of product will be needed, and
the student may be interested even at this stage in reading Chapter XIII of the
author's Advanced Calculus, or Appendix A.
6 MECHANICS
Since sin (180 A) = sin 4, we can state the result in the
following form. Let three forces, P, Q, and E, acting on a par-
ticle, be in equilibrium. Denote the angles between
the forces, as indicated, by p, q, e. Then Equa-
tion (1) represents a necessary condition for equilib-
rium. Conversely, this condition is sufficient.
We thus obtain a convenient solution in all cases
except the one in which the magnitudes of the forces,
but no angles, are given. Here, the Law of Cosines *
FIG. 9 gives one angle, and then a second angle can be
computed by the Law of Sines.
Example 1. Forces of 4, 5, and 6 are in equilibrium. Find
the angles between them.
First, solve the problem graphically, measuring the angles.
Next, apply the Law of Cosines :
42 = 52 + 62 _ 2 X 5 X 6 X cos p,
cos <p
=f , <p
= 41 23'.
5 4
sin i sin <p
5V7
'
55 46'.
16
Example 2. A
40 Ib. weight rests on a smooth horizontal
cylinder and is kept from slipping by a cord that passes over the
cylinder and carries a 10 Ib. weight at its other
end. Find the position of equilibrium.
The cord is assumed weightless, and since it
EXERCISES
1. Forces of 7, 8, and 9 pounds keep a particle at rest, Find
the angles they make with one another.
2. Forces of 51.42, 63.81, and 71.93 grs. keep a particle at rest.
What angle do the first two forces make with each other? Find
the other angles.
3. A weightless string passes over two smooth pegs at the same
level and carries weights of and P
at its ends. P
In the middle,
there knotted a weight W. What
is
2
,
*
~r A n
,
Y= Y + F + t 2 + Yn .
Xk = =
(2) 2)
t-i
0, 2Y
t=i
k 0.
EXERCISE
A4 Ib. acted on by three forces, all of which lie in the
weight is
the vertical, and forces of 8 Ibs. and 12 Ibs. on the other side of
the vertical and making angles of 20 with the upward vertical
and 15 with the downward vertical respectively. Find the force
that will keep the system at rest.
n n^ n
being positive when the component has the sense of the positive
axis of coordinates,and negative when tho sense is the opposite.
In solving problems in equilibrium it is frequently simpler to single
out the components that have one sense along the line in question
and equate their sum, each being taken as positive, to the sum
of the components in the opposite direction, each of these being
taken as positive, also. The method will be illustrated by the
examples in friction of the next paragraph.
fastened to the brick, and let the string be pulled horizontally with
a force F just sufficient to move the brick. Then the law of
physics is that A
F = pR, R\
r
J
-p^
where R (here, the weight of the brick) is p 15
the normal* pressure of the table on the
brick, arid ju (the coefficient of friction) is a constant for the two
surfaces in contact. Thus, if a second brick were placed on top
of the first, R
would be doubled, and so would F.
We can state the law of friction generally by saying: When
two surfaces are in contact and one is just on the point of slipping
* Normal means, at right angles to the surface in question. The normal to a
surface at a point is the line perpendicular to the tangent plane of the surface at
the point in question.
10 MECHANICS
over the other, the tangential force F due to friction is propor-
tional to the normal pressure R between the surfaces, or
n tan X.
W sin X, W cos X.
And now the forces acting up the plane (i.e. the components
directed up the plane) must equal the forces down the plane, or
F = W sin X ;
and the forces normal to the .plane and upward must equal the
forcesnormal to the plane and downward, or
R = W cos X.
Hence
F
R
-
sin X
cosX
= an x^
,
But
F = /J2.
Consequently
n tan X.
STATICS OF A PARTICLE 11
slipping. Then friction acts up the plane, and the forces which
produce equilibrium are those indicated. Hence
'F+W
50 sin 30
W
50 cos 30
Fio. 17 Fia. 18
F + W = 50 sin 30 = 25,
R = 50 cos 30 = 25 V3,
and F = %R.
It follows, then, that
~
W=6 o
25 = 17.8 Ibs.
If,now, W
is slightly increased, the 50 Ib. weight will obviously
17.8 ^ W g 32.2.
EXERCISES
1. If the cylinder of Example 2, 3, is rough, /z = ,
find the
total range of equilibrium. Ans. 4 49' g g 23 44'.
25 cos 6 = R,
25 sin + R = 50.
50 the equation :
Va + 2
62
or
(5) cos (0 a)
=
Va + 2
b2
* The student should observe carefully the trigonometric technique set forth
in this paragraph, not merely because equations of this type are important in them-
selves, but because the practical value of a working knowledge of trigonometry
is not confined to solving numerical triangles. Of far greater scope and importance
in practice are the purely analytical reductions to other trigonometric foi-ms, and
the solution of trigonometric equations. That is one of the reasons why the
harder examples at the end of the chapter are valuable. They not only give
needed practice in formulating mathematically physical data they require also ;
(6)
z2 + 2/
2 = 1,
whose coordinates are
x = __JL__ ' y
7
b
~ __
The angle from the positive axis of x to the radius drawn to this
point is a. Thus we have a graphical determination of a. It is
not necessary to compute the coordinates accurately, but merely
to observe inwhich quadrant the point lies, so as to know which
root of the equation for tan a to take. Thus if b > 0, a must be
an angle of the first or second quadrant. Finally, if c/Va 2 b2 +
is numerically greater than the equation has no solution.
1,
cos (6
- 9 280 -
- 9 28' = 70 48'.
sin 2 -f cos
2 = 1,
we should be led to a quadratic equation in the sine or cosine. This equation will,
have four roots between
in general, and 360, and three of them must be excluded.
Moreover, the actual computation by this method is more laborious.
14 MECHANICS
be called into play, or pR. It seems plausible that such points lie
to the right of the critical point but this conclusion is not im-
;
Mfi-
n --
the amount
50-25 sin
g
available, or
or
if
- --
F < nR
<^50-25
5
b
sin
or if .
cos (0
- 9 28
7
) < -='
EXERCISES
Solve the same problem if the plane is inclined at an angle
1.
of 15 with the horizon, and the vertical plane through the weight
and the pulley is at right angles to the rough plane.
Ana. 43 33' g0g 69 24'.
dx
a
s-. cos x + b sin x
6. Solve the equation :
EXERCISES ON CHAPTER I*
1. A
rope runs through a block, to which another rope is
string, attached to the weight, passes over the edge of the bowl
and carries a weight P at its other end. Find the position of
equilibrium.
12. Solve the same problem for a parabolic bowl, the rim being
at the level of the focus.
13. One end of a string is made fast to a peg at A. The string
passes over a smooth peg at J?, at the same level as A, and carries a
weight P
at its free end. A smooth heavy bead, of weight W, can
slide on the string. Find the position of equilibrium and the
pressure on the peg at B.
14. A bead weighing W Ibs. can slide on a smooth vertical
circle of radius a. To the bead is attached a string that passes
STATICS OF A PARTICLE 17
17. A
weightless bead
*
can slide on a smooth wire in the form
of an ellipse whose plane is vertical. A string is knotted to the
bead and passes over two smooth pegs at the foci, which are at
the same horizontal height. Weights of P and Q iro attached f
P tan <p
= (P + W} tan 6, sin + sin <p
= 1,
where 0, <p are the angles the segments of the string make with
the vertical.
* Questions of this type may be objected to on the ground that a force must act
on mass, and so there is no sense in speaking of forces which act on a masslcss ring.
But if the ring has minute mass, the difficulty is removed. The problem may be
thought of, then, as referring to a heavy bead, whose weight is just supported by
a vertical string. Since the weight of the bead now has no influence on the position
of equilibrium, the mass of the bead may be taken as very small, and so, physically
negligible.
18 MECHANICS
21. If, in the preceding question, P= W y show that = 21 55',
cos e4 - 3 - - 2
|
cos e
I (i ~) cos e
hence
9 cos 4 6 - 6 cos 3 6 - 8\ cos 2 8 + 6X cos 6 - X = 0,
where _ (W - P
2 2
X ~ )
TP
23. Show that if, in the preceding question, P= 5 and
TF = 10,
= 25 8', and find the reaction on the peg.
24. an inextensible string a in. long is made fast to
One end of
a peg A and at the other end is knotted a weight W. A second
string, attached to W, passes over a smooth peg at B, distant
b in. from A and at the same level, and carries a weight P at its
other end. Find the position of equilibrium.
^ -
If P = Wj how far below the level of the pegs will the first
26. A
bridge of simple type is suggested by the figure. In
designing such a structure, the stiffness of the members at a point
A B
J_
FIG. 26
* The
following four problems are given only in outline, and the student thus
has the opportunity of filling in reasonable numerical data and formulating a clean-
cut question. It is not necessary that he respond to all the problems; but he
should demand of himself that he develop a number of them and supplement these
by others of like kind which he finds of his own initiative in everyday life. For,
imagination is one of the highest of the intellectual gifts, and too much effort
cannot be spent in developing it.
STATICS OF A PARTICLE 19
FRICTION
drawn to the bead and the vertical is 16, find whether the bead
will slip when released.
tan i - P
~"
~ W ' 1 tan ? -
" W~P ' 1
2 P+ W /z' 2 W+P n'
Find the other positions of equilibrium, and discuss the
case P = W.
38. Cast iron rings weighing 1 Ib. each can slide on a rough
horizontal rod, M = -J. A string 6 ft. long is attached to each of
these beads and carries a smooth bead weighing 5 Ibs. How far
apart can the two beads on the rod be placed, if the system is to
remain at rest when released ?
39. An elastic string 6 ft. long, obeying Hooke's Law, is stretched
to a length of 6 ft. 6 in. by a force of 20 Ibs. The ends of the
string are made fast at two points 6 ft. apart and on the same
level. A weight of 4
Ibs. is attached to the mid-point of the string
cot = 120 (1
- cos 0).
a 5 ton weight, without the derrick's being pulled over, the dis-
tance from the stone to the derrick being 120 ft.?
CHAPTER II
(1) R =P+ Q,
Moreover,
(3) a +6= c.
(4) R = Q - P,
having the sense of Q, its line of action cutting AC produced in
the pointB determined by the equations :
(5) aP = bQ,
(6) a = b + c.
EXERCISES
1. A
10 ton truck passes over a bridge that is 450 ft. long.
When is one-third of the way over, how much of the
the truck
load goes to one end of the bridge, and how much to the other
end? Ans. 6f tons to the nearer end.
2. A weight of 200 Ibs. is to be raised by a lever 6 ft. long, the
fulcrum being at one end of the lever, and the weight distant 9 in.
from the fulcrum. What force at the other end is needed, if the
weight of the lever is negligible?
3. A two baskets of pottery by a pole 6 ft. long.
coolie carries
If one basket weighs 50 pounds and the other, 70 pounds, how far
arc the ends of the pole from his shoulder?
2. Analytic Formulation ;
n Forces. Suppose that n parallel
forces act. Then two, which
are not equal and opposite, can be
replaced by their resultant, and this in turn combined with a third
one of the given forces, until the number has been reduced to two.
These will in general have a resultant, but, in particular, may form
a couple or be in equilibrium. Thus the problem could be solved
piecemeal in any given case.
An explicit analytic solution can be obtained as follows. Begin
with n = 2 and denote the forces by P l and P2 Moreover, let P 1 .
arid
tities,
AT
P
one direction
2be taken as algebraic quan-
being positive if they act in
T
line
VI*
if
perpendicular*
they act
P 1
and
-
XssQ
--
3J=1
Fl(J
p\
33
1
** *2
(l) R = P, +P
provided P +P ^
l 2 0. Its line of action has the coordinate:
* An oblique direction could be used, but in the absence of any need for such a
generalization, the orthogonal direction is more concrete.
24 MECHANICS
Suppose, first, that P t and P 2 are both positive. Then, by 1,
R = P, +P 2
where
a = x x 1? o = x2 x,
where |
x |
means the numerical or absolute value of x. Thus
|
- 3 = 3, 3 = 3. Moreover,
| | |
let xl < x2 .
Then, by 1,
P and P2 have a resultant,
l
12 = -?! +P 2,
a = x x lt b = x xtl
and hence, from 1, (5) :
(x
- x { ) (- P t)
= (x
- x2 ) P 2.
have a resultant,
R = P! ^
---- + Pn ,
3, ,
m -
1. By hypothesis,
+ + Pm * 0.
Pi
'
Pi + + Pm-, * 0, '
R' = P + + Pw - l
- -
lf
"
/I r * '
w #w~i
x ~
-*
_/ 3*i 'T~ i
/>! + -TP^;
'
R' + Pm = 1\ + - - -
+ Pm * 0,
R = R' + Pm = P + 1
- -
+ Pw ,
_
X
R'x' + Pm x m P lgl + +Pw a,
#' + PM A + --+P
But this result contradicts the assumption that the theorem is
(4) P, + - -
+ Pn = 0,
-
Pn * 0.
Thei1 P + + Pn-! ^ 0,
l
-
' '
-/ = P! %l H~ "T Pn-i ^n~i
26 MECHANICS
R' +P n
= 0, or R = - Pn
f
_ PI X \ I
' '
~T 1 n-i X n -i
Xn
fr> n
y
_^ P, X l H
-----h Pn-i gn-i
~
arulonlyif
PI + . . .
+ P. = 0.
Pi X l + ----h Pn X n = 0.
- m l
x + l
-
+ mn xn
m + l
-
+ mn
STATICS OF A RIGID BODY 27
(sec below) and its coordinates, (x, ?/), are given by Equations
(1) and
/ _
~ *KI y\ H----+ yn >n
(2)
9x fl
y
'
n^ + .-.+m.
For, let the plane of the particles be vertical, the axis of x being
horizontal. Then the system is acted on by n parallel forces,
whose lines of action cut the axis of x at right angles in the points
%!,'', x ny and is determined in position by Equa-
their resultant
tion (1). On
rotating the plane through a right angle and repeat-
ing the reasoning, Equation (2) is obtained.
The centre of gravity of any material system, made up of par-
ticles and line, surface, and volume distributions, is defined as a
point, (7, the parts of the system be rigidly connected
such that, if
m + l
.
+ mn
.
c) Solid cone : x = f h.
d) Conical surface : x f A.
e) Triangle : Intersection of the medians.
28 MECHANICS
4. Moment of a Force. Let F be a force lying in a given
plane, and let be a point of the plane. By the moment of F about
is meant the product of the force by the distance from of its
line of action, or hF. A moment may furthermore be defined as
an algebraic quantity, being taken as positive when it tends to turn
the body in one direction (chosen arbitrarily as the positive
direction), and negative in the other case. Finally, if lies on
(1) xY-yX.
Proof. Let the equation of the line of action of F be written
in Hesse's Normal Form :
x cos a + y sin a = h.
/
O
i FIG. 34 FIG. 35
a = - + 2rr.
Hence
cos a = sin 6, sin a = cos 0,
and since
X = F cos 6, Y = F sin 6,
the proof is complete.
STATICS OF A RIGID BODY 29
,- a --- Bt
- + |.
Hence
cos a = sin 9, sin a = cos 6.
xY 1 yX l
and xY 2 yX 2
.
The components of the resultant force are seen to take the form :
X +X l 2 and Y + F
l 2, and its moment is
is seen to be :
Suppose first that the forces of the one couple are parallel to the
forces of the other couple. Then, by proper choice of the axis
of x, we can represent the couples as indicated, where
P. + P.-O. 0<P,;
P.I ,
J
I l . P +P =
3 4 0, 0<P 3 ;
p "
*\ P, x^xt + h, 0<h;
J
*4
= * + 3 J, < l >
Fxo.36
P, +P -P -P = 2 3 4
and
Pr^l _|_p-r
1 I
*
2 ** /
2
_ P *
,.
3 "^S
_ /:>
*
4
/r
**/
4
(P
^JTj -f P2 / ^2
-4- X W P
-I
i
It (P
V.^3
-\- I
P4/}
-* '
P
r*'4 4- ^3 7
^ = fl
VJ>
P
now appear as and P.
STATICS OF A RIGID BODY 31
forces obviously are in equilibrium, for all four are equal in magni-
of y', and two equal and opposite forces at along the axis of
y. These forces constitute a resultant couple, whose moment is
the sum of the moments of the given couple.
This last statement is at fault in one particular. It may happen
that the second couple is equal and opposite to the first, and then
the resultant forces both vanish. In order that this case may not
cause an exception, we extend the notion of couple to include a
nil couple i.e. a couple whose forces are both zero, or whose forces
:
and such a couple is a force equal to the given force, but trans-
ferred to a new line of action, parallel to the old line.
It may happen that the resultant force vanishes, but the result-
ant couple does not. For equilibrium, it is necessary and suffi-
cient that both the resultant force and the resultant couple vanish.
This condition can, with the help of the Remark at the close of
5, be expressed in the following form.
i) they are such as would keep a particle at rest if they all acted at
a point; and
ii) the sum of the moments of the forces about a point (one point
is enoughj and it may be chosen anywhere) of the plane is zero.
X r
= 0,
r=l
= 0.
2) (XrYr- yrXr)
S = nR.
Thus
/i
= 2-V3 = 0.27.
EXERCISES
1. If in the example just discussed the wall is smooth, but the
floor is rough, and if /* = ,
find all positions of equilibrium.
The theorem has already been proved for the case that the
easily seen that the resultant of the two forces at A and the result-
ant of the two forces at B form a new couple.
Finally, the vector representations of these three couples are
three vectors perpendicular respectively to the three planes of
the couples, equal in length to the forces of the couples, and so
36 MECHANICS
oriented as to give the same figure yielded by three of the forces,
duce at two forces that are equal and opposite to the force Fk .
(1) R = F! + + Fn ,
resultant couple, C, is :
(2) C = C, + - - -
+ Cn ,
C = C, + C2 ,
and only if both the resultant force and the resultant couple
vanish.
For completeness it is necessary to show that the resultant
force, R, together with its line of action, and the resultant couple,
C t,
are uniquely determined. For it is conceivable that a differ-
STATICS OF A RIGID BODY 37
ent choice, O', of the point O might have led to a different result.
Now, the vector R is uniquely determined by (1), and so is the
same but C depends on the choice of 0', and so C l
in each case ;
(2) M = x a y b z c
X Y Z
L = (y-V)Z- (z-e)Y
(3) M = (z -c)X - (x- a)Z
N = (x-a)Y - (y-
P:(x,y,z)
r = r' - rj,
r' = x i + y j + z k,
FIG. 45
* The student should read 3 of Appendix A. This, together with the mere
definitions that have gone before, is all of Vector Analysis which he will need for
the present.
38 MECHANICS
Vector Representation of a Couple. Let a couple consist of two
forces, F! and F 2 :
F, +F = 2 0,
(4) C = r XP lf
(5) C = r, X P! + r2 XF 2,
hence
r X F! = T! X F!
- r2 X Fj = T! X Fj + r2 XF 2
.
forces in space. Let P be any point of space, and let equal and
opposite forces, F* and FA ,
k = 1, , n, be applied at P.
Consider the n forces F^ - -
,
Fn which act at P. Their re-
sultant is
R = F! + -
+ Fn .
-
C* = r* X F* r X F*.
force,
(6) R = F, + + Fn ,
(7) C = 5) r* X F* - r X R.
fc-i
STATICS OF A RIGID BODY 39
(8) RXC = 0, R* 0.
Let
(9) iXF^Li
(10) R = Fj + Zk.
Thus
(11) L =
RX (Li + M + tf k) = R X
j (r X R),
or:
(12)
Y + zZ = YN - ZM
(13) zZ) = ZL
- XN
- Z (xX zZ) = XM
- YL
One of these equations may become
illusory through the vanish-
ing of all but some two always define intersecting
the coefficients ;
Hence
(15) $
YN -ZM >?
= ZL-XN ZM- YL
2
ft
40 MECHANICS
Thus we have found one point of tt*e resultant axis, and the
direction of the axis is that of R. The resultant couple is given
by (7), where
(16) r = i + Tjj + fk.
Wrench. A
wrench is defined as two forces, acting at arbitrary
points; moreover, neither force shall vanish, and their lines of
action shall be skew.
Let the forces be F*, acting at (xk, y^ Zk), k = 1, 2. The reader
will do well to compute the resultant force, axis, and couple.
EXERCISE
Let F! and F2 be two forces, the sum of whose moments about
a point is 0. Show that F t F2 and , ,
lie in a plane.
r
L^r By the moment of F about L is meant :
FlG 46 -
M = hY a,
r X F =-
k (r X F) = hY.
Hence
(1) M = {a-(rXF))a.
More generally, r may be any vector drawn from a point of L
to a point of L'. We have thus arrived at the following result.
STATICS OF A RIGID BODY 41
M = {a- (r X F)}a,
where r is any vector drawn from a point of L to a point of the line
(2) 2 (y k Zk - zk Yk ) =0, 2 (z k X -
k xk Zk ) = 0,
ii) The sum of the vector moments of the given forces about
an arbitrary line of space is 0.
f
ii ) The sum of the vector moments of the given forces about
each of three particular non-complanar lines is 0.
42 MECHANICS
Necessary and Sufficient Conditions. It is important for clear-
ness to analyse these conditions further, as to whether they are
necessary or sufficient or both.
Condition i), regarded as a necessary condition, is broadest when
is taken as any point of space. But Condition ii) is sufficient
if it holds for the lines through just one particular point 0, the
condition (1) being fulfilled.
EXERCISES
1. Show that Condition i) is sufficient for equilibrium.
and since they have the same sense, they have a resultant not 0.
Let the axis of z be vertical and directed downward. Then
the resultant is a force directed downward and of magnitude
n
(1) R = Wj g + + mn g = g 5) m k,
The nearest point of the resultant axis to the origin has the coordi-
nates given by (15), 10 :
i j k
X ij 2
mk a mk p mk y S nik y
(m + n) cot if/
m cot 6 n cot <p,
where \l/
is the angle this line makes with the
segment n.
The proof is immediate. Project the sides
of the triangle on this line, produced :
(m + n) cos \(/
= a cos 6 cos <p t
FIG. 47 m b n
sin i sin0' sn sn <p
Example 1. A
uniform rod of length 2a is held by a string
of length 21 attached to one end of the rod arid to a peg in a
smooth vertical wall, the other end of the rod
resting against the wall. Find all the positions
of equilibrium.
The three forces of W, T, and R must pass
through a point, and this must be the mid-point
of the string. Hence, applying the above trigo-
nometric theorem to either of the triangles ABO
or ABC, we have :
or:
2
4 cos
cos 2 = <p
1+3 cos 2
<p
cos <p
=
But a and I are not unrestricted, for < cos ? < 1. Hence
2a cot <p
= a cot 6 a tan \,
-+ - -+
Hence
72
Z
2
sin 2 = - - r-rr = 2
sin 2
+1
:
A :
: ;
2
Z
4 cos 2 <p + 4/i cos <p sin <p + (1 + /j
2
) sin 2 <p
=
This equation, in turn, could be reduced to a quartic in sin <p
or cos <p but such procedure would be bad technique. Rather,
;
let
2 cos 2 <p
= 1 + cos 2y?, 2 cos <p sin v?
= sin 2<p,
p =
2
2 sin 1 cos 2<p.
EXERCISE
1. Complete the study of Example 2, i) computing A, B, C,
and ii) finding when the rod is just on the point of slipping up.
- - 5 - M
2
(K2
J
;
2. If a = 1, I = If, p,
= 0.1, find all positions of equilibrium.
weighing 5 Ibs. walks along the top of the gate, and the gate
drops. The disconcerted cat springs off. It is observed from
her tracks in the snow that she reached a point 2 ft. from the
end of the gate. The distance between the hinges is 2^- ft.,
and the centre of gravity of the gate is 5 ft. from the vertical
line through the hinges. If the gate weighs 100 Ibs., what is
the bowl and the rim of the bowl in contact with the rod. Find
the position of equilibrium.
Ans. cos 6 = a -f V32r
2
+a 2
,
or
where the radius of the bowl is r, the distance of the centre
of gravity of the rod from its lower end is a, and the inclina-
tion of the rod to the horizon, 0, provided a < 2r.
7. A uniform rod rests with one end on a rough floor and the
other end on a smooth plane inclined to the horizon at an angle a.
Find all positions of equilibrium.
8. The same problem where both floor and plane are rough.
9. A picture hangs on a wall. Formulate the problem of
equilibrium when the wall is smooth, and solve it.
10. The same question where the wall is rough.
11. A smooth rod rests with one end against a vertical wall,
a peg distant h from the wall supporting the rod. Find the
position of equilibrium. [ft
Ans. cos = \* a
12. The same problem where the wall is rough, the peg being
smooth. Find all positions of equilibrium.
14. A
plank 8 long is stood up against a carpenter's work-
ft.
either the floor or the bench and the plank is If the plank .
16. A uniform rod 2 ft. long rests with one end on a rough
table. To the other end of the rod is attached a string 1 ft. long,
made fast to a peg 2 ft. above the table. Find all positions of
equilibrium.
Ans. One system of limiting positions is given for ju
= 2
by the equations :
cot <p
= 2 + 2 cot 0, 2 cos 6 + cos <p
= 2.
Approximations.
17. A water tower is
100 ft. high and 100 ft. in diameter. Find
approximately the tension in the plates near the base.
handle, what is the condition that the drawer will stick fast ?
CHAPTER III
MOTION OF A PARTICLE
Rectilinear Motion.* Tube simplest case of motion of mat-
1.
ter under the action of force isHhat in which a rigid body moves
without rotation, each point of the body describing a right line,
and the forces that act being resolved along that line. Consider,
for example, a train of cars, and neglect the rotation of the wheels
and axles. The train is moved by the draw-bar pull of the loco-
motive, and the motion is resisted by the friction of the tracks
and the wind pressure. Obviously, it is only the components of
the forces parallel to the tracks that count, and the problem of
Dynamics, or Kinetics, as it is more specifically called, is to
determine the relation between the forces and the motion; or,
if one will: Given the forces, to find the distance traversed
as a function of the time.
A more conventional example, coming nearer to possible experi-
mentation in the laboratory, would be that of a block of iron
* The student must notobliged to finish this chapter before going on. What
feel
is needed is a thorough the treatment of the early problems by the present
drill in
methods, for these are the general methods of Mechanics, to inculcate which is
a prime object of this book. Elementary text-books in Physics sometimes write
down three equations :
* = ^at
2
, v at, t?
2 = 2as,
and give an unconscionable number of problems to be solved by this device. The
pedagogy of this procedure is totally wrong, since it replaces ideas by a rule of
thumb, and even this rule is badly chosen, since it disguises, instead of revealing,
the mechanical intuition. Now, a feeling for Mechanics is the great object to
be obtained, and the habits of thought which promote such intuition are, fortu-
nately, cultivated by just the same mathematical treatment which applies in the
more advanced parts of Mechanics. It is a happy circumstance that here there
is no conflict, but the closest union, between the physics of the subject and the
mathematical analysis.
A thorough study of 1-12 through working each problem by the present general
methods is most important.
Moreover, 22 should here be included with, of
course, the definition of vector acceleration given in 16 and the statement of
Newton's Second Law in 17. The student should then turn to Chapter IV, the
most revealing chapter in the whole elementary part of the book, and study it ic
all detail. The remaining sections of the present chapter should be read casually
at an early stage, so as not to impede progress. Ultimately, they are important ;
but they are most useful when the student comes to recognize their importance
through his experience gathered from the later work above referred to.
49
50 MECHANICS
placed on a table and drawn along by cords, so applied that the
block does not rotate and that each point of it describes a right
line with varying velocity.
It is clear that a block of platinum having the same mass,
i.e. containing the same amount of matter, if acted on by the
same would move just like the block of iron, if the two were
forces,
started side by side from rest or with the same initial velocities.
We can conceive physical substances of still greater density, and
the same would be true. On compressing the given amount of
matter into smaller and ever smaller volume, we are led to the
idea of a particle, or material point, i.e. a geometrical point, to
which the property of mass is attached. This conception has
the advantage that such a particle would move exactly as the
actual body does if acted on by the same forces; but we need
say nothing about rotation, since this idea does not enter when
we consider only particles. Moreover, there is no doubt about
where the forces are applied they must be applied at the one
point, the particle.
(2) t.
(4) ma oc /.
(5) ma = X/,
unit of mass, the foot the unit of length, the second the unit of time,
and the pound the gravitational unit of force, then X has the
value 32(or, more precisely, 32.2), and Newton's Second Law of
Motion becomes here :
A,) m
52 MECHANICS
In the decimal system, the gramme being the unit of mass, the
and the
centimetre the unit of length, the second the unit of time,
A 2)
mf = 981
t
/.
dv .
A
A,)
\
m di =f -
U
^
x/2 *
/!\
(6) a =
(7\
dv _ dsdv
~
( '
dt di ds
(8) . - .*.
dv dz s dv
Here we have
m- X 2000 = 400,000 Ibs.,
200
= 32X8000,
400,000^
or =
dt 25
V == Tjngt ~f~ C .
FlG 52
the force will be ^m lbs.,f and
*
dv 16
V =
ds ~-5>
or v dv = /-ds.
2
" 16 n
rr
Hence T>
= s + C.
.
^-
= + C, C = 450,
^!
and so v 2 = 900 - ^s.
When the stone stops, v = 0, and we have
= 900 - 3s, s = 141 ft.
*
EXERCISES
1. If the train of Example 1 was moving at the rate of 4 m.
3. Show
that, the coefficient of friction between a sprinter's
if
and (t'i) the coordinate used, as s or x, etc. He should not try to adapt such formulas
of Elementary Physics as
v = at, a = a2 , vz = 2as
to present purposes. For, tho object of these simple exercises is to prepare the way
for applications in which the force ia not constant, and here the formulas just cited
do not hold.
MOTION OF A PARTICLE 55
i.e. the constant force which would produce the same velocity in
the same time.
6. In the preceding problem, assume that the given speed is
acquired after running 200 feet. Find the time required and
the average force.
6. A train weighing 500 tons and running at the rate of 30 miles
X
This multiplier reduces to unity when we take a = e. Hence the definition :
The absolute (natural) base of logarithms is that base which makes the multiplier
logo e in the above equation equal to unity.
56 MECHANICS
which gravity attracts the unit of mass. Then the force, measured
in absolute units, with which gravity attracts a body of units m
of mass will be mg. Newton's Second Law A 3 ) gives for this case :
dv , dv
ds
s = %gt* + K, K = 0,
and we have the law for freely falling bodies deduced directly
from Newton's Second Law of Motion, the hypothesis being
merely that the force of gravity is constant. Substituting in
the last equation the observed values s = S, t = T, we get :
9
= 28
?*'
v = gt, s = %gt
2
.
dv
2gs.
MOTION OF A PARTICLE 57
o
= + <?>
= g (sin 7 - cos 7) +v .
B) s = \g (sin 7 - ^ cos 7) P +V Q t,
dv -
v-r = f
g (sin 7
.
/x
,
cos 7),
do
%v
2 = g (sin 7 M cos 7) s + K,
^ = + X,
t;
2 = 20 (sin 7 /z cos 7) s + #o-
EXERCISES
1. If, in the example discussed in the text, the body is pro-
jected up the plane, find how far it will go up.
2. Determine the time it takes the body in Question 1 to
reach the highest point.
3. Obtain the usual formulas for the motion of a body pro-
jected vertically :
v 2 = 2gs + vl or = + vl
2gs ;
v = gt + V Q or =- +V gt Q ;
8 = ot* + v t or =-tf* + M.
58 MECHANICS
4. On moon a pound weighs only one-sixth
the surface of the
as much on the surface of the earth. If a mouse can jump
as
up 1 foot on the surface of the earth, how high could she jump
on the surface of the moon? Compare the time she is in the air
in the two cases.
(1) T oc s, or T = ks,
T = 12 when s = .
(2) T = 24s.
T = k's,
we have :
120 = k' \ or k' = 240,
as given
When the string is
T =
stretched 1
2 pounds.
in., s = ^ and the tension
Formula (3), on the
by (2) is 2, i.e.
l
f
,
the stretched length; the coefficient being Young's Mod- E
ulus. For a given string, E/l = k is constant, and V I = s is
variable.
* It is easy to check an answer in any numerical case. The student has only
to ask himself the question: "Have I expressed force in pounds, or have I my
expressed it in terms of half-ounces?" Just as five dollars is expressed by the
number 5 when we use the dollar as the unit, but by the number 500 when we
use the cent, so, generally, the smaller the unit, the larger the number which expresses
a given quantity.
60 MECHANICS
EXERCISES
1. An clastic string is stretched 2 in. by a force of .3 Ibs. Find
the tension (a) in pounds; (b) in poundals, when it is stretched
s ft. Ans. (a) T = 18s ; (6) T = 180s.
2. When the string of Question 1 is stretched 4 in., what is
and it is found that the length of the string has thus been doubled.
Find the tension in the string when it is stretched s ft.
Ans. fs Ibs. ; ^s poundals.
5. A Problem of Motion. One end of the string considered
in the text of 4 is fastened to a peg at a point of a smooth
horizontal table ;
a weight of 3 Ibs. is attached to the other end
of the string and released from rest on the table with the string
.stretched one foot. How fast will the weight be moving when
the string becomes slack?
The weight evidently describes a straight line from the starting
point, A, toward the peg 0, and we wish to know its velocity
when it has reached a point B, one foot from A.
The solution is based on Newton's Second Law of Motion.
-- --
It is
-
convenient here to take as the coordinate, not the distance
T-,
<
riG. 54
KA
'
-j
AP
that the particle has travelled at
any
The
but its distance s from B.
instant,
which acts is the tension
force
i.i
of the string; measured in absolute
units it is 240rs. Since it tends to decrease s, it is negative.
Hence Newton's Law becomes :
(1)
___
fJ2 a tit)
To integrate this equation, replace -^ by its value v -=- :
(2)
MOTION OF A PARTICLE 61
Hence v dv =
I v dv = 8g I sds,
(3)
*
EXERCISES
1.The weight in the problem just discussed is projected from
B along the table in the direction of OB produced with a velocity
of 8 ft. per sec. Find how far it will go before it begins to return.
Ans. Newton's equation is the same as before, and the
integral, (3), is the same ;
but initially s
= and v = 8.
3^=-24g8 + i-3g,
and the answer is : 4 Vl5 = 15.49 ft. per sec.
3. Solve the problem of Question 1, for a rough table, M = T-
Ans. The required distance is the positive root of the
equation 16s 2 +s 4 = 0, or s = .4698 ft., or about
5fin.
* In the following exercises and examples, it will be convenient to take the value
of g as exactly 32 when English units are used. Begin each exercise by drawing a
figure showing the coordinate used, and mark the forces in red ink.
62 MECHANICS
4. Find where the weight in Question 2 will come to rest
if the string, after becoming slack, does not get in the way.
below A.
(5) v = = V8g Vl s 2.
-^
(6)
(7) I6dt =-
If we measure the time from the instant when the body was
released at A, then t = arid s = 1 are the initial values which
determine C :
=- sin- 1 1 + C, C = ~
* In evaluating the above integral we might equally well have used the formula
Vl - 2
(9) W= cos- 1 =
|,
t =
J2
= .09818 sec.
EXERCISES
1. Show that if the body, in the case just discussed, had been
released from rest at any other distance from the peg, the string
being stretched, the time to the point at which the string becomes
slack would have been the same.
2. Show that it takes the body twice as long to cover the first
TrA/^ = .9618
Ans. = \^K
t
*
04 J/ v/6s s
-',
2
total time,
^ sec.
any point of the hole, to the distance of the stone from the
centre: ~ ,
F oc r, or F = kr.
mg = kR or k = ~^,
/
and
F = ^r
R
As the coordinate of the stone we will take its distance, r,
from the centre of the earth. Then Newton's Second Law gives
us:
/i\
(1) m d*r = -R
^ mg r -
w
(2) <*L=-L r
dp R >
a 4?
<&c cte 2
Integrating, we have
66 MECHANICS
Cd ( --^.Cr^
~ r
Jdt\dt R dt
or
~_ -- -
dt R R
To determine (7, observe that initially, i.e. when the stone was
at Ay r = R and the velocity, dr/dt, = 0. Hence
0=~|ft2 + c, or C = !# 2
.
7 it
Thus finally :
<3 >
(I)'- 1 <'-">
At the centre of the earth, r = 0, and (dr/dt) = gR. If we 2
take the radius of the earth as 4000 miles, then R = 4000 X 5280,
g
= 32, and the velocity is about 26,000 ft. a sec., or approxi-
mately 5 miles a second.
The stone keeps on with diminishing speed and comes to rest
for an instant when r = J?, i.e. it just reaches the other side
'"
dr
dt =- \ F====^.
9 VR* - r
or t = cctfh
1 ~+ C.
^\|
I
Initially, t = and r = 72 ;
thus (7 = 0, and
(4) t = \
*
cos- 1 ~, or r = R cos (^ VB) '
<7 /t \ A//
MOTION OF A PARTICLE 67
*
-'
X
A\
'
**
~_ % ~
'
~dfi
B)
C) x = h cos ntj
^ 27T
FIG. 56
EXERCISES
1. Two
strings like the one described in the text of 4 are
fastened, one end of each, to two pegs, A and B, on a smooth
horizontal table, the distance AB being double the length of
either string, and the other end of each string is made fast to
a 3 Ib. weight, which is placed at 0, the mid-point of AB. Thus
each string is taut, but not stretched. The weight being moved
to a point C between and A and then released from rest, show
that oscillates with simple harmonic motion.
it Find the velocity
with which it passes and the period of the oscillation. It is
assumed that the string which is slack in no wise interferes with
or influences the motion.
Ans. The differential equation which dominates the motion
d2 x =
is
ffi 256z, where x denotes the displacement of the
at
3 Ib. weight ;
hence the motion is simple harmonic motion.
The required velocity is numerically 16h, where h denotes
the maximum displacement. The period is 27T/16
=
.3927 sec.
2. Work the same problem for two strings like the one of
Question 4, 4, and a 2 Ib. weight.
3. Show
that the motion of Example 7, 5, is simple harmonic
motion, and find the period.
round wooden spar, so that, when left to itself, the spar floats
vertically in water. A ten-kilogramme weight having become
accidentally detached, the spar is seen to oscillate with a period
of 4 seconds. The radius of the spar is 10 centimetres. Find
the sum of the weights of the spar and attached iron. Through
what distance docs the spar oscillate ?
Ans. (a) About 125 kilogrammes ; (6) 0.64 metre.
El
A
*-?
Since F = mg when r = JR, the radius of the earth,
X
mq = -F^ and
,
vr = mgR*
^
70 MECHANICS
Newton's Second Law of Motion here takes on the form :
z 2
d r _
m ~dT* ~ mgR
~7*~'
Hence
drd 2 r 2gR 2
dr d/dr\* = 2gR*dr
r
2
dt dt 2 r2 dt' di\dt) r dt
Initially dr/dt
= and T = Z ;
hence
o.' + c, c~
so, no matter how great I is, V can never quite equal V2gR.
Here g = 32, R = 4000 X 5280, and hence the velocity in ques-
tion is about 36,000 feet, or 7 miles, a second.
This solution neglects the retarding effect of the atmosphere;
but as the atmosphere is very rare at a height of 50 miles from
the earth's surface, the result is reliable down to a point com-
paratively near the earth.
In qrder to find the time it would take the stone to fall, con-
sider the equation derived from (2) :
Hence 2
MOTION OF A PARTICLE 71
, Vl r rdr
and
dr
Vlr - r2
= V - r + ~ sin-
lr
2
Thus t =
fji
Initially t = and r = I :
Finally, then,
fr-H + s s-sm-
For purposes of computation, a better form of this equation
is the following :
(3)
EXERCISES*
1. If the earth had no atmosphere, with what velocity would
6. the earth and the moon were held at rest at their present
If
mean distance apart, and a stone were placed between them at
the point of no force and then slightly displaced toward the earth,
with what velocity would it reach the earth ?
*Or, more generally, to one and the same point P of a rigid or deformable
material body.
MOTION OF A PARTICLE 73
xt +i Xk = Azfc,
the work, ATF*, done by the force in displacing the particle through
the fc-th interval ought, in order to correspond to the general
physical conception of work, to lie between the quantities
(6) W = Jf(x)dx,
a
Here, F
to be taken as a negative
is number equal numerically
to the intensity of the force.
Thus (7) is seen to hold in whichever direction the force acts,
provided that a < 6. Will (7) still hold if 6 < a? It will.
There are in all four possible cases :
i) ++ ii) Hi) H h
In cases i) and ii) the force overcomes resistance, and positive
work is done. In cases in) and iv) the force is overcome, and
negative work is done. Hence (7) holds in all cases.
It is now easy to see how the definition of work should be
laid down when F varies in any continuous manner. The consider-
ations are precisely similar to those which led to Equation (6),
and that same equation is the final result in this, the most general,
case : &
W = Cf(x) dx.
a
O A P B
~~
FIG. 59 T = \
= I/\ x x r
and
,
W
TT7
\jdx
= ,
j
I xdx =
,
-==-
MOTION OF A PARTICLE 75
This is the work done on the wire by the force that stretches it.
Ifthe wire contracts, the work done by the wire on the body to
which its end P is attached will be
Xa 2
21 21
/(->!)*<"
EXERCISES
1. In the problem of 7 compute the work done by the earth
on the particle when the latter reaches the centre.
2. A particle of mass m moves down an inclined plane. Show
that the work done on it by the component of gravity down the
plane is the same as the work done by gravity on the particle
when it descends vertically a distance equal to the change in
level which the particle undergoes.
3. A particle is attracted toward a point by a force which
is inversely proportional to the square of the distance from 0.
How much work is done on the particle when it moves from a
distance a to a distance b along a right line through 01
4. If the earth and the moon were stopped in their courses
mv 2 '
Let the particle move under the action of any force F which
varies continuously: F = /(s). Then Newton's Second Law
can be written in the form :
dv f/ ^
fi
5 -/().
Hence
mvdv = f(s)ds.
76 MECHANICS
Integrate this equation between the limits a and 6, denoting
the corresponding values of v by v l and v2 :
/ mv dv = I f(s) ds.
mv 2 Pa
~2~ n
The right-hand side is, by definition, the work W done on the
particle by the force F. Hence
(1)
i
EXERCISES
Work the Exercises of 5, 7, 8, so far as possible, by the
Method of Work and Energy.
1 = 3c, c = i,
and,t
(2)
'
= $8.
^2 <?
= 32/,
(3)
*<y
* The word substance here
may be too narrow in its connotations, for we want
a word that will include every measurable quantity, from the length of a light-
wave to the wheat crop of the world. Such a woid obviously does not exist,
and so we agree to use substance in this sense as a terminus tcchnicus.
t It would seem paradoxical to say that the same line has a length of 6 when
the foot is the unit, and a length of 2 when the yard is the unit. But it must bo
remembered that the length is a function of two variables, the unit being one of
them. The attempt is sometimes made to meet the apparent difficulty by saying
"3 ft. = 1 yd." But this makes confusion worse confounded; for 3 = 1 is not
true, while on the other hand to try to introduce "concrete numbers," like 3 ft.,
10 Ibs., 5 sees., into mathematics, is not feasible. To try to change units in this
way leads to blunders and wrong numerical results. There is only one kind of
number in elementary mathematics. To attempt to qualify it as abstract, is to
qualify that which is unique. The denominate attribute (3 ft., 10 Ibs., etc.) is part
of the physical thing conceived it does not pertain to the mathematical counterpart,
;
Next,
m = f
km,
_
- v - m' -
m m
i
I j.
ft X 9000
ZOOO, lc
lc
20Q()
,
2QO()
Similarly,
i' _ J
? = $
60' 2.20'
Thus
2
60 2 2
m ,d"^ -_ s'
md^ s
2000*X3
XT
*j = __L_\'/
A;-
220
The left-hand sides of these equations are equal by (4). On
equating the right-hand sides and dividing by (3) we find :
60 2 X'
X' = 422.4.*
2000 X 3 2.20 X 32'
EXERCISES
1. the units of length, time, and mass are respectively a mile,
If
a day, and a ton, compute the absolute unit of force in pounds.
2.the acceleration of gravity
If is 981 in the c.g.s. system,
compute g in the English system.
3. If the acceleration of gravity is 32.2 in the English system,
compute g in the c.g.s. system.
* More bo tabulated as
precisely, the result should :
since the data, namely, X - 32, are correct only to two significant figures.
MOTION OF A PARTICLE 79
each term must have the same dimension. For, such an equation
remains true, no matter what the units of mass, length, and time
may be; and if two terms had different dimensions in any one
of the fundamental quantities (mass, length, time), a change
of units would lead to a new equation not in general equivalent
to the old one.
This principle affords a useful check on computation. Thus,
if an ellipse is given by the equation :
V = 7ra6
2
.
EXERCISES
Determine the dimension of each of the following quantities:
1. Kinetic energy. Ans. ML T~2 2
.
6. Couples. Ans. ML 2
T' 2 .
11. In Question 10, what is the answer when the wind re-
sistance is taken per square foot of surface exposed ?
one hundred per cent literal, can be used to yield a correct result,
when R= 4000.
16. Examine each equation in 8 as to whether the Check
of Dimensions is applicable.
MOTION OF A PARTICLE 81
(1) R = av,
(2) R= cv*
graph.
Problem 1. A man is rowing in still water at the rate of 3 miles
an hour, when he ships his oars. Determine the subsequent
motion of the boat.
Here Newton's Second Law gives us :
/o\ dv
(3)
m^=-av.
TT j*
= m dv
Hence at ,
a v
/A\ A
t
m
=- i
VQ
(4)
log-^
where V Q is the initial velocity, nearj^^Bft. a sec.
a*
= i *>o
log--, or
m v v
Hence
_at
(5) 9 v = VQ e .
It might appear from (5) that the boat would never come to
rest, but would move more and more slowly, since
_at
lim e~ = 0.
= 00
dv
mv~r = av,
ds
and consequently :
=
(6)
o-^.
Hence, even if the above law of resistance held up to the limit,
the boat would not travel an infinite distance, but would ap-
proach a point distant
feet from the starting point, the distance traversed thus being
proportional to the initial momentum.
Finally, to get a relation between s and t, integrate (5) :
ds -?
/>7\
(7) 8 = ffiVn /1
-(l
- e
~m\
m
).
From this result is also evident that the boat will never cover
a distance of S ft. while the above approximation lasts.
EXERCISE
If the man and Ibs. and if a steady
the boat together weigh 300
force of 3 Ibs. is maintain a speed of 3 miles
just sufficient to
an hour in still water, show that when the boat has gone 20 ft.,
the speed has fallen off by a little less than a mile an hour.
R= ct;
2
.
MOTION OF A PARTICLE 83
The forces which act are i) the force of gravity, mg, downward,
2
and if) the resistance of the air, cv upward. As the coordinate
,
of the particle we will take the distance AP, Figure 60, which
it has fallen. Then, Newton's Second Law becomes :
dv
= mg , A
cv 2 .
m-jj
cv 2
TT
Hence
dv
v~r = mg
as m ,
cv*
mv dv
mg cv*'
s =- log (mg
- cv 2 ) + <7,
and thus
/ON
(8)
v 7
finally
s
W m
= 5-
2c
i
log
m#
<7- cv 2
00
-t'
FIG. 60
2
mg cv
(9)
v y ^gg-^-^g
When s increases indefinitely, the last term approaches as
its limit, arid_hence the velocity v can never exceed (or quite
equal) v Vmg/c ft. a sec. This is known as the limiting velocity.
It is independent of the height and also of the initial velocity, and
EXERCISES
1. Work Problem 2, taking as the coordinate of the rain drop
its height above the ground.
2. Find the time in terms of the velocity and the
velocity in
terms of the time in Problem 2.
3. Show that, if a charge of shot be fired vertically upward,
it will return with a velocity about 3 times that of rain drops
84 MECHANICS
of thesame size and that if it be fired directly downward from
;
or
dv
m _-/().
ff ,
/() = av
b
.
MOTION OF A PARTICLE 85
EXERCISES
Work out the relations between v and s, and those between
v and I, ifthe only force acting is the resistance of the medium,
which is represented by the formula :
1. R = a + bv + cv\ 2. R = ~-~V --
3. R = av*.
7 + dv
speeds as at low ones and that the resistance of the air is pro-
portional to the square of the velocity, find by how much the
speed of the train will have dropped off in running half a mile
if the steam is cut off with the train at full speed.
during the descent and the time required to reach the bottom.
Find the limit of velocity that could be acquired by a tobog-
gan under the given conditions if the hill were of infinite length.
Ans. (a) 68 ft. a sec. ; (b) 30 sees. ; (c) 74 ft. a sec.
is 12 ft. If no air escaped from the well, how far would the
elevator sink in? What would be the maximum weight of a
man of 170 pounds? Given that the pressure and the volume
of air when compressed without gain or loss of heat follow the
law :
l 4l
-
=
pv const.,
(1) *-/(*), 2/
MOTION OF A PARTICLE 87
~P**' As n =
r
hm - = rhm = Dts v,
n
numerically.
A =O At AJ-=O At
dx ~
__ dy "
__
dz
'
Vx Vv Vz
~Tt' ~dt'
EXERCISES
1. Show that, if polar coordinates in the plane are used, the
component velocities along and orthogonal to the radius vector
are respectively :
dr dB
where 6 and <p denote respectively the co-latitude and the longi-
tude.
3. A point moves in space. Show that
dr dB
(1) Av = v' - v,
cf. Fig. 65, p. 96. The average vector acceleration is defined as the
vector
Av
AT
When A approaches 0, the average vector acceleration approaches
a limiting value, and this limiting vector is defined as the vector
acceleration of the point :
/ \ r
= hm Av
(a)
Ar=o At
puted. For, the components of the vector (1) along the axes
are respectively :
v'x vx v'y
vy v'z vz
' '
AJ AJ AZ
A/ = *** A/ = &t A/ = At
or:
d'2 z
v = *-
--a
Let s be the arc, measured in the sense of the motion ;
and let
ds
Then r' is a unit vector lying along the tangent and directed
in the sense of the motion. Furthermore,
= r
ds
dv dsdr
= ,
and *-__
Hence
EXERCISES
1. A point describes a circle with constant velocity. Show
that the vector acceleration is normal to the path and directed
toward the centre of the circle, and that its magnitude is
2
"
,
or or2 r.
re 6 \
W>-J7Z
= X,
(2)
d?z
t
y"72 o tYllfi
m = T = N
where p = l//c.
MOTION OF A PARTICLE 93
EXERCISE
Show that
(r' x r") F = 0.
-(z'x")Y+(x'y")Z = 0,
where
y'z"-z'y", etc.
and
ff
/
1
dx f fr
JU ~; . JU etc.
ds
m dt
2
(2)
FIG. 64
= - = -
^ v Q sin a 0tf , y V / sin a
94 MECHANICS
Eliminating t we get :
*>
- #?,
U
ff
'
~2T~
Transforming to a set of parallel axes through A, we have :
x = x
1
+ a?!, y = y' + y lt
V ' * 2 2 COS 2
&UQ LUo
n,
Cc
+
.
"
~~2g~ 2i~ 27'
The result is independent of the angle of elevation a, and so
it appears that all the parabolas traced out by projectiles leaving
with the same velocity have their directrices at the same level,
the distance of this level above being the height to which the
projectile would rise if shot perpendicularly upward.
EXERCISES
1. Show that the range on the horizontal is
R = sin 2a,
j/
~
g'
The height of the directrix above is half this latter range.
2. A
projectile launched with a velocity of V Q ft. a sec. and
is
is to hit a mark at the same level and within range. Show that
there are two possible angles of elevation and that one is as much
greater than 45 as the other is less.
*
~
~g 1 + sin
MOTION OF A PARTICLE 95
Gin.
If a man puts the shot 46 ft. and the shot leaves his hand at
a height of 6 ft. 3 in. above the ground, find the velocity with
which he launches it, assuming that the angle of elevation a is
the most advantageous one. Am. v = 35.87.
(1)
mv 2 =
N.
P
give one for the plane case without the use of vector methods.
Geometric Proof. Compute the components of the vector
acceleration along the tangent and along the normal. Let <p
96 MECHANICS
be the angle which the tangent has turned through in passing
from P to P' Then the component of Av along the tangent
.
will be
f
v cos tp v = (t; + Av) cos <p v
= lim~r, = PP'
K p lim
PP'
f
v cos & v
= Av 1 cos (p
-
TT -r- COS <f>
V
At At At
/,. AZA/V \ ~
f lim = D v.
Mlimcos^J t
1 cos y? __
1 cos <p jp_
As
At <p As At
The first factor approaches 0, and the second and third factors
remain finite, since each approaches a limit. Hence the limit
of the right hand side is 0.
v sm <p
At
This can be written as
; sin <p _
(?_
As
v
<p As At'
MOTION OF A PARTICLE 97
EXERCISE
Use the present geometric method to obtain the formulas :
1 d
where a r ,
ot e denote respectively the components of the vector
acceleration along and perpendicular to the radius vector.
and since s = 1 0,
2g . de
~~T: 77^ ~T~ sin u -jr.
dt dt 2 I at
= cos + C,
(2)
^ = ^(cos0-cos).
The velocityin the path at the lowest point
is times the angular velocity for
I = 0, or
V20Z (1 cos a), and is the same that would have been acquired
if the bob had fallen freely under the force of gravity through the
MECHANICS
same difference in level. Equation (2) is virtually the Integral
of Energy.
If we attempt to obtain the time by integrating Equation (2),
we are led to the equation :
de
20 J Vcos cos a
(3)
-~
_ g
dz s
s = 4a sin r,
d*s g
FIG. 67
,
we have -j-z2
= -~ s.
at 4a
dx
(1)
dt* ds
Hence
\A:(xQ ,y )
_
~ 9g
2
~ds~dt
But the velocity that a body falling freely from rest a distance of
x x attains is expressed by precisely the same formula.
In the more general case that the bead passes the point A
with a velocity v we have :
eg
= 2gx<> + C,
(3) t>
2 - eg
= 2g (x
- z ).
it reaches the end of the wire or comes to a point A' at the same
v* >^L level as A
In the latter case it will in gen-
eral just rise to the point A and
1
then retrace
itspath back to A. But if the tangent to the
FIG. 69
curve at A
'
is horizontal, the bead may
approach A 1
as a limiting position without ever reaching it.
EXERCISES
1. A bead slides on a smooth vertical circle. It is projected
from the lowest point with a velocity equal to that which it would
acquire in falling from rest from the highest point. Show that
it will approach the highest point as a limit which it will never
reach.
= "% + RD
.
*>
~dT*
d*y_
(4) ~* ~ D "'
where Rx Rv R
, , t are the components of the reaction of the wire R
along the axes. Since R
is normal to the curve, we have :
But
2 2
/efo\ (dy\
= .
/dz\*
(&) + U) + b)
2
"2
^ m d (v ) = mg dx.
2
(7)
/o\
(8)
mvZ ---mv a
= "V (*
/ - \
*<>)
-7j 2^
This precisely the Equation of Energy.
is It could have been
written downat the start from the Principle of Work and Energy.
It is the generalization of (2) for space curves.
EXERCISE
A bead slides on a smooth wire in the form of a helix, axis
vertical. Determine the reaction of the wire in magnitude and
direction.
N ,r
= mv =
2
- r
mco 2r.
9 FIG. 70
"
Why, then, the term centri/u^aif force" the force that "flees
the centre"? The explanation is a confusion of ideas. If the
mass is held in its path by a string fastened to a peg at the centre,
0, does not the string tug at in the direction OP away from the
centre and is not this force exerted by the particle in its attempt,
or tendency, to fly away from the centre? The answer to the
first question is, of course, "Yes." Now one of the standard
methods of the sophists is to begin with a question on a non-
controversial point, conceded without opposition in their favor,
and then to confuse the issue in their second question "and
is not this force exerted by the particle?"
102 MECHANICS
Matter cannot exert force, for a force is a push or a pull, and
matter can neither push nor pull it is inert. The particle does
;
not pull on the string, the string pulls on the particle. But even
this statement will be accepted only half-heartedly, if at all,
by people who have not yet grasped the basic idea of the science
of Mechanics the study of the motion of matter under the
action of forces. What comes first is a material system solid
aere. In the present case there are two systems, each of which
can be isolated (1) the particle (2) whatever the peg is attached
:
;
belong to the only true church ; they subscribe to the Red Cross
drive but they have no place in the Temple of Science ; they
profane it.
FIG. 71
where m = ^, v = 600, r = 5 ;
or
600 2
4500
16X5 ;
T cos a. ~ mg.
Hence
v2 44 2
Since the distance between the rails is 4 ft. 8 in., it follows that
the outer rail must be raised 3.42 in.
104 MECHANICS
EXERCISES
1. A weighing 4 oz. is attached to a string which
particle
passes through a small hole, 0, in a smooth table and carries
a weight W
at its other end. If the first weight is projected
the string being taut and the part below the table vertical, how
great must W
be, that the 4 oz. weight may describe a circular
path? Ans. 9 Ibs. 12 oz.
2. A
boy on a bicycle rounds a corner on a curve of 60 ft.
radius at the rate of 10 m. an hour, and his bicycle slips out from
under him. What is the greatest value p, could have had ?
Ans. Not quite |.
3. A conical pendulum is like a simple pendulum, only it is
west at the same rate, what is the increase in her apparent weight ?
to the surfaceon the side toward the sun and evenly distributed
as regards a cross-section by a plane at right angles to them,
show that they would have to be as close together as blades of
grass. It is assumed that their other ends are guided near the
earth's surface.
forces of gravitation.
or
" X '
dx g
It follows, then, that
0)
v-$*.
Thus it appears that the free surface is a paraboloid of revolution
106 MECHANICS
To Graduate the Cup. It is easily shown that the volume of
a segment of a paraboloid of revolution is always half the volume
of the circumscribing cylinder. If, then, we mark the level of
the oil when it is at rest, the height, h, to which it rises above this
level when it is in motion will just equal the depth, h, of the
lowest point of the surface below this point. From (1) it follows,
then, that if a denotes the radius of the cup,
or
EXERCISES
1. A tomato can 4 in. in diameter is filled with water and
sealed up. It is placed on a revolving table and caused to rotate
about its axis, which is vertical, at the rate of 30 rotations a sec.
Find the pressure on the top of the can.
Ans. The weight of a column of water 4 ft. high (nearly)
and standing on top of the can.
2. How great is the tendency of the can to rip along the seam?
+ 2
(mco r)
2 =
and it would make an angle <p with the downward vertical, where
tan <p
= wV
MOTION OF A PARTICLE 107
tion, we might think a large round cup, 500 ft. across at the
of
g
it follows that
~ '
32
co = T4T (nearly), or .36.
27T
-
jT,
or
T = = 17 sees.
0>
2
(0.004r) ,
and upon the rim of the cup, this would amount to mgV2, or
41 per cent greater than gravity on the fixed surface of the earth
roughly, two-fifths more. A movie actress who was main-
taining her weight in Hollywood, would tip the scales at,
well, how much ?
What we have said applies, however, only to bodies that are
at rest in the field. When a body moves, still other forces enter,
and these be considered in the chapter on Relative Motion.
will
Nevertheless, we can
describe the motion of a projectile directly,
since it would be a parabola in the fixed space we started with.
Imagine a tennis court laid out with its centre at the lowest
point of the bowl. Lob the ball from the back line to the back
line, and watch the slice !
108 MECHANICS
One may reasonably inquire concerning the engineering prob-
lems of the construction. There will be a tendency of the cup
to burst to fly apart, due to the "centrifugal force." Can it
(1 >
d.(r*\ = '
rdt\ dt)
(2) r
^ ==h '
A = / r*dd,
~~
FIG. 74 dt dt
MOTION OF A PARTICLE 109
and hence
(3) A=ft(-g,
or, equal areas are swept out in equal times.
We have tacitly assumed that h 5* 0. If h = 0, then (2)
reduces to dd = 0, and the path is a straight line through 0.
Work and Energy. The kinetic energy of the particle is
W
W nw^_m(W
2
"
2
+r W <
=
~ 2
i
2
2 2
(6) TF = Cfidr.
Hence
I 1
_
~ /
2
7> mh*J J?
=-
(8)
ll^,
Thus
(9) ^
The Law of Energy, as expressed in the form of Equation (7),
here becomes :
110 MECHANICS
where X = m/x, and C is a constant depending on the initial condi-
tions.
The form of this equation suggests a simplification consisting
in substituting for r its reciprocal :
(11) u =
J-
Thus (10) becomes :
/io\ = 2/4 ~
(12)
^ 5 + i 9 -
S;
+
.
C'.
* - u ~
?
Then (13) goes over into :
(14)
*
dr 2
+^- B".
(15) x = B cos (0
- 7),
+ x* - u '
de*
*
and the only solution of this differential equation is
x = 0.
If S * 2
0, then (14) yields :
d8 =
where, however, the two signs are not necessarily the same.
But in all cases this last equation leads to (15).*
We set out to integrate Equation (12), and we have arrived
at the result :
(\7\
1 } r = -
-
-
M 1 ecos(0 7)
The The path of the particle is given by Equation (17).
Orbit.
This the equation of a conic referred to a focus as pole and
is
M (1
- e*)
2
h* . A
-
Md-e 2
)' M Vf=T'
The distance between the foci is
(19) c =
The area of the ellipse is
(20) TTOfc =
(1
- e )1
2
Th
The periodic time T is connected with the area A by the rela-
tion:
A = pr.
Hence
(21) T2 = 47r
2 --
(22) h = vr sin ^,
/f>O\ _
r* /'-j //j \\
Hence
Furthermore,
du = fie
-
.
2
sin (Q -y).
dO h'
Since
du __
1 dr __
i dr __
v cos \f/
'
d6 r 2 dO h dt h
we have initially :
7 =
h h M
From (22),
(28) e
MOTION OF A PARTICLE 113
and this equation contains the interesting result that the orbit
will be the following conic :
i) ellipse, if v% < ;
ii) parabola, if v\
= ;
-=4 =
(30)
an 2
h2 MO.
A2 = via*.
(31) v* =
Jj.
EXERCISES
1. Show that if
= u =
, ,
r2 -rr h and -,
dt r'
then
114 MECHANICS
2. been assumed that the orbit is a plane curve. Prove
It has
this tobe the case by means of a constraint, consisting of a smooth
plane through 0, the point of projection, and the tangent to the
path at that point. Use Newton's Equations, 17, (2), and
show that the force of the constraint is 0.
/ = K
(r + r ')2>
mr = m'r'.
Hence
r
, _ m+ m'
r'
~m'
and so
= f-m-
\m + ml
(
-
'
\22
\
,)
,
m'.
r^
T
dt
= h ft '
u =
1 e cos (07) , p = const.,
,
P
d*u = 1
W* + U
.
p>
and
mh*l
K~ r*
p
The force is, therefore, an attractive force, inversely proportional
to the square of the distance from the centre, when r lies between
its extreme values for this ellipse. But an arbitrary range of
values, < a < 7 < 0, can be included in such an ellipse, and so
the result is general.
EXERCISE
Show that if the path is an ellipse with the centre of force at
the centre, the force is proportional to the distance from the
centre.
T2 = 47T 2 -,
M
win
EXERCISES
1. Show that the first of the equations (1), 25 :
dt* dt 2
* For a more detailed treatment cf. Appell, Mecanique rationnette, vol. 1, 3d ed.,
1909, 229 et seq.
MOTION OF A PARTICLE 117
.
r*
r
dt
u = ~ + B cos (B 7).
2. Prove that
h
v -,
P
where p denotes the distance from to the tangent to the path.
Is it an ellipse?
remain within the earth's orbit, assumed circular, for more than
\
- 1-th part of a year, or nearly 76 days.
(2
8. A shell is describing an elliptical orbit under a central
attractive force. Prove that, if itexplodes, the pieces will
all
assume the physical law that action and reaction are equal and
opposite; i.e. that the vector F21
>2/2)
is equal and opposite to the vec-
tor F 12 or
,
Fi2 + F21 = 0.
*i2 +X = 21 0, F + 721 =
12 0.
j. =
__
X + Xn + X
1 1
There are in all three such pairs of equations, those in x being the
following :
d*x d2 x
d2 x
= -^3 +
~dfi
120
DYNAMICS OF A RIGID BODY 121
a X\ a Xn Ci X-i
m *~dP + m*~W + m *^W = r
+ X>2 + * r
.
,
-* . *r .
Xl 3
*
m *
"eft
2
^
"^ "* 2
^
^ft 2
"^ "* 3
~eft^
~~ * * ^~ * ? ~"~ * 3 '
x = mn x n 2
mn
(2)
= mn yn
V m +1 + wn
If we denote the total mass by M, then
!
rn k xk =
Hence we have :
d*x_
a'Xkk
~M ~
l*l/
MTUT- ~jfLi
H/w
V
2*
fc j^
(3)
122 MECHANICS
These equations are precisely Newton's Second Law of Motion
for a particle ofmass Af, acted on by the given external forces,
each transferred to the particle. We can state the result as
follows.
(4) *
Remark.
=
There
^ is
,
A)
(1)
Now, ~
a '
s - ld*s I
I dt*'
G *
Moreover, from the laws of freely
'
Opl
falling
bodies,
(2) OP (s
- 1) + F- gpl.
Hence
(3) F = gps + pv
2
,
or F = 3gps.
This means that the pressure of the chain on the table is always
just three times the weight of that part of the chain which has
already come to rest on the table.
It appears, then, that F is made up of two parts, i) the pres-
sure gps on the table, of that part of the chain already at rest;
and ii) a pressure
(4) P= pv*,
PA== ov *
DYNAMICS OF A RIGID BODY 125
The Crew on the River. The crew is out for practice. Ob-
serve the cut-water of the shell and describe how it moves, and
why it moves as it does. What system do you decide to isolate ?
the shell? or the shell, oars, and crew?
EXERCISES
1. If a man were placed on a perfectly smooth table, how
could he get off?
2. If a shell were fired from a gun on the moon and exploded
in its flight, what could yo.usay about the motion of the pieces ?
3. A goose is nailed up in an air tight box which rests on plat-
form scales. The goose flies up. Will the scales register more or
less or the same?
4. A pail filled with water is placed on some scales. A cork
isheld submerged by a string tied to the bottom of the pail.
The string breaks. Do the scales register more or less or the
same?
6. A man, standing in the stern of a row boat at rest, walks
forward to the prow. What can you say about the motion of
the boat?
6. When the man stops at the prow of the boat, boat and man
will be moving forward with a small velocity. Explain why.
7. A uniform heavy string is laid over a smooth
flexible
compute the velocity of the string when the upper end reaches
the cylinder, correct to three significant figures.
9. Find how long it takes the upper end of the string to reach
the cylinder.
(1) xY-yX.
Consider a system of particles acted
on by any external forces whatever,
and interacting on one another by
FIG. 78
forces that are equal and opposite, but
are now assumed each time to lie in the lino joining the two par-
ticles in question. Moreover, the particles shall lie in a fixed
m l
r
and compute its value from the equations in question, namely,
(x, Y -
l Vl X,) + (x, Y -
12 Vl X 12 ) '+ (x, Yn - y, Xn ).
* It is important that the student do this, and do it neatly, and not merely gaze
at the three equations printed in 1 and try to imagine the three not printed.
He should write out the full equation derived below from these, neatly on a single
line, and then write the other two under this one.
DYNAMICS OF A RIGID BODY 127
THEOREM OF MOMENTS:
B) m k
d*Xk - yA).
Sill
_
-7T,
dt $--
We observe next that, in all generality, by mere differentia-
(3)
<L ( x d-i
y
4*\- x
d*y _
di\ ~di dt)~ d?
* Cf. the footnote, I.
128 MECHANICS
and we proceed to compute the parenthesis by means of Equa-
tions (1) and (2). We find :
dy dx ~dO
d ( dyk dx k \ _ d*8
Xk -~
yk rt
dt\ dt ~dt) ~dT
for r k does not change with the time, and so drk /dt = 0. Hence
But m k rk = Mh y
(6)
I^ =-MghsmO.
2
^\ d2 6 g . .
* Moments of inertia for such bodies as interest us here are treated in the
Author's Introduction to the Calculus, p. 323.
DYNAMICS OF A RIGID BODY 129
or
l =
m
(8 2 ) I =
j,
where 7 = MW,
k denoting the radius of gyration.
More precisely, what we mean by the last statement is this.
Let a simple pendulum be supported at 0, let its length be k*/h,
and let it be placed alongside the rod, the bob being at a point
distant Z from O. If now both be released from rest at the same
instant, they will oscillate side by side, though not touching each
other.
The Actual Rod. As n grows larger and larger, the massless
rod weighted with the n particles comes nearer and nearer to
the actual rod, dynamically. This is not a mathematical state-
ment. It expresses our feeling from physics for the situation
our intuition. And whon we say that the motion of the actual
so
rod is the limit approached by the motion of the auxiliary rod,
we are stating a new physical postulate. The result is, that the
actual rod oscillates like a simple pendulum of length
2 2
fc_ -iL -?/
h # 3
EXERCISES
Apply the method set forth in the text, introducing each time
an auxiliary set of particles, and proceeding to the limit. Do
not try short cuts by attempting to use in part the result of the
exercise worked in the text.
Then
FIG. 81
for example, the line through and
the centre of gravity, G, of the
particles, and denote the angle it makes with the axis of x by 0.
Then
where ct k varies with fc, but is constant as regards the time. Hence
d0k =
" de d 2 Bk =
~~
d*0
dt dt' dP dt*
(2) 2)
" mkgyk = -02) mky*'
The last sum has the value My, whore the coordinates of G
are denoted by (x, y). Let the distance from to G be h. Then
y = h sin
and (2) becomes
(3)
- M0/i sin 6.
obviously an
unnecessary restriction. Conceive a block of
EXERCISES
Find the length of the equivalent simple pendulum when the
compound pendulum is one of the following.
7 = Mfc 2
^A
:
(6) I = -
,_* + *
x
(7) x* - Ix + Jk
2 = 0,
DYNAMICS OF A RIGID BODY 133
(8) h* - Ih +k = 2
0.
^ -_i - I
j,
ft _ ~
* 2
THEOREM. The locus of the points 0, for which the time of oscil-
lation is the same consists of two concentric circles with their centre
y
EXERCISES
1. Draw two concentric circles about G, of radii h and k*/h.
Show that the length, Z, of the equivalent simple pendulum cor-
of radius k.
(2) (M + m) a 2 -^ + Ma -^ 2
Thus B) becomes :
2
dt
mga'
(5)
dt 2 I + (2M + m) a
2
(6) s =
I + (2M + m) a 2
~ 0.
II
10 Ibs., and that 5 Ib. weights are fastened to the two ends of
the string. What will be the acceleration due to a 1 oz. rider?
2. Work the case in which the wheel
is a hoop, i.e. a uniform
=
B) / -jT2 5J Moments about 0.
* This problem is more difficult than the others, and is essentially a problem in
the Calculus ; cf. the author's Advanced Calculus, Chapter 14, 8.
DYNAMICS OF A RIGID BODY 137
EXERCISES
1. A
turn table consisting of a uniform circular disc is free
to rotate without friction about its centre. man walks along A
the rim of the table. Find the ratio of the angle turned through
by the table to the angle described by the man, if man and table
start from rest.
cf. Example 5.
coordinates, (x, y, z) and let the second axis be the axis of z'
;
/ =
2m k (x,? + 2
2/* ), 7 =
2 w* W + yi
2
).
Since
x = x' + x, y = y' + y,
it follows that
=
) 2
k .
2x mk xi + 2y <k yi-
The last two terms vanish because 0' is the centre of gravity,
and hence
=
2m k x't o, 2
It remains merely to interpret the terms that are left, and
thus the theorem is proved for a system of particles.
lim
2 *(**
2
+ 2/t )
2 = Km
n-oo j7 n = oc
or
since these limits are by definition the moments of inertia for the
continuous distribution.
DYNAMICS OF A RIGID BODY 139
= Ma
2
(2) T = 2*
of gravity were held fast and the same forces acted on the lamina as
those applied in the actual case y
=
(1) ^"77/2 S Moments about (?,
140 MECHANICS
where I denotes the moment of inertia about the centre of gravity, G;
6 is the angle that a line fixed in the lamina makes with a line fixed
in the plane, and the right-hand side is the sum of the moments of
the forces about G.
origin is at G. Then
(2) x = + a, y = 77 + y,
and
>k *
= 0, ] m k rj k
= 0,
Remembering that
dt\dt
we see that Equation B), 3, here becomes :
Because
xk = & + x, yk = ri k + y,
2 -
7 fen rjkXk) +
Since
DYNAMICS OF A RIGID BODY 141
it follows that
(5) *
-. FIQ 84
(2) R = Mg cos a,
Let y denote the distance of the centre of gravity from the plane.
* A pipe, the thickness of which is negligible, when placed on the plane with its
axis horizontal, would move in the same way. The two problems are dynamically
identical.
142 MECHANICS
Then
(4)
= aF, I = Ma 2 .
/-^
Since there is no slipping,
(5) 5 = a0,
Ma -JTJ = Mg sin a F,
(6)
= aF.
Ma*<^
On eliminating F between these equations, we find :
or
(8)
^= |
sina .
Hence it appears that the centre of the hoop moves down the
plane with just half the acceleration it would have if the plane
were smooth.
Equation (2) appears to have played no part in the solution.
But we have assumed that there is no slipping, and so F cannot
be greater than pR :
(9) F ^ R.
F ^y s n a *
'
2
DYNAMICS OF A RIGID BODY 143
Mg sin a ^ . .
2-5 M^ cos a,
(10) tan a ^ 2M .
EXERCISES
1. Show that, if the hoop be released from rest,
v ~ sm a,
= gt .
s = at*
~- sm a, .
v2 = 0s sin a.
2. Show furthermore that
2
at at
= TJ-
.
sin a, ^ = T- sin a,
.
<&& 4<z
W 2 *L gj na
a
3. Solve the problem studied in the text for a sphere. Show
that
d*6 5g d*s 50
= . .
^5 ^ am a, ^J^Bina.
tan a g JJLI.
__
dt*
(1) F
R = Mg FIG. 85
d* e
tv\
(2) 7r = nw
^ IT
(3) F = MB.
From Equations (1), (2), and (3) it appears that
/A \ &S s= -
(4.) i
r
'
~
d< 2 2a
. f v = v - ngt, s = v t-
,g
\ t>
2 = vl
-
and
Thus as the ball advances, its centre moves more and more
slowly, while the speed of rotation steadily increases. Finally,
pure rolling will set in. This takes place when the velocity of
the point of the ball in contact with the table is nil. Now, the
velocity of this point of the ball is made up of two velocities,
namely, i) the velocity of translation, or the velocity the point
would have if the ball were not rotating, i.e. v, as given by (50;
and ii) the velocity due to rotation, or the velocity the point
would have if the ball were spinning about its centre, thought of
as at rest. The latter is a velocity of ao> in the direction opposite
to the motion of the centre, and is given by (5 2 ). Thus the ve-
locity forward of the point of the ball in contact with the table is
(6) v ow.
(7) v oo> = 0.
DYNAMICS OF A RIGID BODY 145
or
Finally,
= =
(11) 1 |>0,
7
! Tla
2'
EXERCISES
1. Solve the same problem in case the table is slightly tipped
and the ball is projected straight down the plane.
s = oQ, and since equations (1) and (2) still hold, we have :
O
FIG. 86
before. But Equation (2) now becomes
(13)
Furthermore,
(14)
Hence
EXERCISES
1. If the centre of the ball was moving initially at the rate of
6 ft. a sec. and if the ball stops after rolling 18 ft., show that
C = IMa.
2. If the initial velocity of the centre was V Q and if the ball
rolled I ft.,show that C is proportional to the initial kinetic energy
and inversely proportional to the distance rolled.
(i) p= &
all the friction now being called into play. On eliminating P
and R, we find :
z
d
- s -
= <7(sin<*
(2)
-= = cos a.
dt 2 a
The integrals of these differential equations can be written
--
down at once. In particular, it is seen that the ratio of s to 6
is constant, if the hoop starts from rest :
s a (sin a u cos a) - -,
- = - '- = ,,
o(tan a cot X
, %
1).
p cos a
DYNAMICS OF A RIGID BODY 147
The last parenthesis has the value 1 when tan a = 2/z, and
is > 1 when a is larger. Thus the motion is one in which a cir-
cle of radius , _
r = a (tan a cot X 1)
(3)
"77/2
= a^ s n *
"~ a^ cos ^ ^ =
FIG. 88 U/t"
- 30 dO
dt
Since
d
it follows that
2
fdO\ 30 .
(dt) =-2H s
The constant of integration, C, is determined by the initial
=- jj
sin a + C.
Zd
Hence, finally,
*> '-(-..-*.)
To find where the ladder will leave the wall. This question is
2 2
RE = M d2 x =
--trr
at 1
/>d
Ma smO-jrz
at
-
2
Ma cos 6*/de\
-77
\at/
[ ) ,
Hence R = when
3 sin 6 2 sin a = 0.
ft
= sin- 1 (f sin a).
(9)
Vsin a sin 6
x = sin 6,
the integral goes over into an Elliptic Integral of the First Kind,
and can be treated by well-known methods; cf. the Author's
Advanced CakuluSj Chapter IX.
(10)
=-aRcosO.
at*
(11) y osinfl.
150 MECHANICS
On eliminating R and y we find :
(12) (fc
2
+a 2
cos 2 0)
^-
(Zf
a 2 sin cos (9
(~f)
\ttf /
=- ag cos 0.
And this can be done here. But in the present case there is a
short cut, due to the special form of the differential equation.
It is observed that, on multiplying the equation through by
(13)
(F +o 2
cos 2 0)
\Gfv /
=-2ag sin 9 + C.
_
~ 2ag sin a
Wl
kick up and leave the table before the centre comes clear down?
The moron certainly cannot answer this objection by physical
intuition.
mathematics sits as judge over the situation.
It is here that
EXERCISES ON CHAPTER IV
(material surface).
3. A billiard ball is set spinning about a horizontal axis and
is released, just touching the cloth of the billiard table. How
far will it go before pure rolling sets in ?
9. A
billiard ball is placed on a billiard table inclined to the
horizontal at an angle a, and is struck full by the cue, so that
it starts off straight down the plane without any initial rotation.
.
12. A plank can rotate about one end, on a smooth horizontal
table. A man, starting from the other end, walks toward the
pivot. Determine the motion.
13. A smooth tube, the weight of which may be neglected,
can turn freely about one end. A rod is placed in the tube and
the system is released from rest with the rod horizontal. Deter-
mine the motion.
14. A spindle consists of two equal discs connected rigidly
with an axle, which is a solid cylinder. The spindle is placed
on a rough horizontal table, and a string is wound round the axle
and carried over a smooth pulley above the edge of the table.
A weight is attached end of the string and the system
to the lower
is released from rest.
Determine the motion.
Consider first the case in which the string leaves the axle from
the top then, the case that the string leaves the axle from the
;
5 above the track and midway between the axles, which are
ft.
weightless string, wound round the rim of the disc, has a weight
of P Ibs. fastened to it. The system is released from rest with
the 1 Ib.weight at the lowest point and the other weight hanging
freely at the same level. How great may P be, if the 1 Ib. weight
is not to be pulled over the top?
19. A billiard ball rolls in a punch bowl. Determine the
motion.
20. A solid sphere is placed on top of a rough cylinder of revolu-
tion, axis horizontal, and slightly displaced, under the action of
gravity. Find where it will leave the cylinder.
21. A
uniform rod is released from rest, inclined at an angle,
with its lower end in contact with a rough horizontal plane.
Will it slip at the start? Determine the motion.
22. A packing box is sliding over an icy side walk. It comes
to bare ground. Will it tip up?
CHAPTER V
KINEMATICS IN TWO DIMENSIONS
1. The Rolling Wheel. When a wheel rolls over a level road
without slipping, the nature of the motion is particularly acces-
sible to our intuition, for the points of the wheel low down move
X
FIG. 90
x = XQ + cos 6 sin 77 0,
y 2/0 + ? sin + y c s 0-
EXERCISE
Express and rj in terms of x and y y i) geometrically, by reading
the result off from the figure ; ii) analytically, by solving Equa-
tions A) for , 77. The formulas are :
f f
= (x
- x ) cos + (y
- 2/0) sin 0,
I rj
= (x XQ) sin + (y y Q ) cos 0.
-^ -, -^-
156 MECHANICS
These derivatives can be computed in terms of the known func-
tions (1), namely, x y Q 0, and of their derivatives, by means
, ,
dx ___
dx Q ,
(2)
dy _ dy* ,
f
dt dt
-(y- X - XQ.
Hence
dx _
~
dt ~dt
dy__ dy, ,
_ dB
dt~~dt +(X
(4) X
.
_dx
~ '
x = etc.
dt*'
Thus the formula for the components of the velocity assumes tho
final form :
* = z - (y
- 2/0) 0>
B) -
y = 2/0 + (* #o) *
that x = and y = 0, or :
= x - (y
- j/ ) 6,
(5)
= y + (*
- z n) 6.
KINEMATICS IN TWO DIMENSIONS 157
C)
Xn
- - - - + A
EXERCISES
1. Show that the coordinates (j, 77 j) of the instantaneous
centre, referred to the moving axes, arc given by the equations :
^o si n 6 ?y n
"
cos &
a
Vx + 2
y
2
ft = ru, < 6.
f *=-&-&)*,
(2)
1 y = (x
- a)6.
= * - (Vi
- 2/oH
= 2/0 + (x l
- x )6,
KINEMATICS IN TWO DIMENSIONS 159
curve, being fixed in space. And now the first cam is allowed
to roll without slippingon the second cam.
Thus a great variety of motions of the lamina can be realized,
and now the remarkable fact is that all motions can be generated
in way, with the single exception of the translations,
this
tives of the second order, and the space centrode is traced out
ii) that the arc of the one curve corresponding to any two
different values of X, namely, Xj and X 2 and the arc of the other;
(1) * = ff(A),
= A(X),
ously :
(2) <<7'(X)
2
+ /i'(X)
2
.
(3)
(4)
KINEMATICS IN TWO DIMENSIONS 161
x = x + % cos 6 TI sin 0,
(5)
y = 2/o + sin 9 + rj cos 0.
Vx
_ dx
~ __
~~ dy
Vy
d\' d\
is tangent to C at P and its length is
vx = u$ cos B Uy sin 0,
(7)
vy = u% sin B + u^ cos B.
tions (8) :
<8)
neous centre in space, and in Equations (8), the (x, y) are the
coordinates of this same point, (x l9 yj. The other quantities
that enter into (8) are the functions (6) that determine the posi-
tion of the moving body; and X = t. Thus Equations (8) go
over into the following :
f *o
- (2/i
- 2/o) * = 0,
1 2/o + (x l
- z ) 6 = 0.
- dy /d0
X = Xn
dX / dX'
(10)
dx a I dB
y = 2/o + -3r
,
/ 3T-
But these conditions are not sufficient, since the functions x and
y thus defined will not in general admit derivatives.
To meet this latter requirement we demand, therefore, that
the functions (6) possess, furthermore, continuous second deriv-
atives. But this is not enough, even if the case that x and y
reduce to constants is excluded (rotation about a fixed point).
It is, however, sufficient when we add the hypothesis of (2),
4, and so demand that
dy / de d de
\
~dx/dx/'
be not both (dO/d\ being, of course, 5^0). In other words,
the equations
d0 d 2 x _ d^cteo _, ^!^o =
"*" U|
dX dX 2 dX 2 dX dX 2 dX
(12)
" "" "~ =
t dX'dX 2 dX*"d\ dX 2 dX
dX
164 MECHANICS
Let the origin, furthermore, be taken at this point (x y ), and ,
d\ 2 ~d\' d\ 2 d\
The second, dy/d\, has the value 0, since C is tangent to the axis
of x at the origin. Hence we infer that
Example. A
billiard ball is projected along a smooth hori-
zontal table with an initial spin about the horizontal diameter
which is perpendicular to the line of motion of the centre. Deter-
mine the two centrodes.
Take the path described by the centre of the ball as the axis of x,
and the centre of the ball as (x y ). Then ,
FIG. 93
x = XQ , y =--.
c/co below the centre of the ball, and the instantaneous centre
is always beneath the centre of the ball. This means that the
ball rolls without slipping on a right line distant c/w below the
centre. Hence the body centrode is a circle of radius c/co about
the centre of the ball.
EXERCISE
A billiard ball is struck full by the cue. Determine the
space centrode and the body centrode during the stage of slipping ;
SC ~~ *C rt ~~ Cv """"
_ 2a _ 2acl
V ~
5 5/ t'
where a denotes the radius of the ball, and c the initial velocity
of its centre. The time that elapses during the stage of slipping
is 2c/7/A<7 seconds. The space centrode meets the billiard table at
the angle
coordinates, are :
2ac 2a
P =- 1
'
t 5
(1) y = sin -
166 MECHANICS
This curve gives satisfactorily the part of the figure not too near
the lines y = 1, but it is tangent to these lines, whereas it
from one which is tangent to the axis of x into one which has a
cusp on the axis, it is enough to replace f(x) by [/(z)]* :
(2) y
expanded with the heat, throwing the shovel onto the other
plate. Then the process was reversed. Thus vibrations like
those of the tea cup arose, and died down.
Vk
= + (& cos
- 77* sin 0) 0.
velocity.
Second Proof. The result may also be obtained by means of
the instantaneous centre, 0. For the motion, so far as the
velocities that enter into the definition of T are concerned, is
one of rotation about 0. Hence
(2) T = i/'ft
2
,
(3) V = Aft.
On substituting this value of /' in (2) and then making u$e of (3),
T takes on the form (1), and this completes the proof.
Generalization. The most general rigid bodies with which we
are concerned are made up of particles and material distribu-
tions spread out continuously along curves, over surfaces, and
throughout regions of space. When such a body rotates about
an. axis, the kinetic energy is defined by Equation (1). We
can state the result in the form :
T
168 MECHANICS
Remark. The formula holds even for the most general case of
motion of any rigid distribution of matter in space. For, such
motion is helical, i.e. due to the composition of two vector fields
of velocity, i) a field corresponding to rotation about an axis;
and ii) a field of translation along that axis cf 12 below.
;
.
EXERCISES
1. A ball rolls down a rough plane without slipping. Deter-
mine the kinetic energy in terms of the velocity of its centre.
instant, form a vector field which coincides with the vector field
arising from the permanent rotation of space about this axis.
We give first a geometrical proof which appeals strongly to
the intuition. The refinements which a critical examination
of the details calls for are best given through a new proof by
vector methods.
Let be a point of the fixed space, distinct from O. If its
Q
velocity then the velocity of every point of the indefinite
is 0,
with their first derivatives, for we are not concerned with dis-
continuous motions.
Pass a great circle, (7, through Q perpendicular to the vector
velocity of P
will also be at right angles to the plane of C and on
the same side of C as the vector at Q. For, since the vector
velocity of Q is at right angles to the chord QP, the vector velocity
of P must lie in the plane through P perpendicular to QP. But
it also lies in the tangent plane to the sphere at P. And now I
to this Exercise.
5. The same for this statement.
(1) Xr;
cf. Appendix A.
Let a system of Cartesian axes (or, y, z) be assumed with as
origin, and let i, j, k be unit vectors along these axes. Write
(2) co 2 k.
Then
i J k
(3)
x y z
KINEMATICS IN TWO DIMENSIONS 171
vx =
(4) Vy
= XUg ZO)
Vg
=
v = () X r,
V= (') X r,
(6) v + V = () X r + (') x r.
Hence
(8) v + v' = {() + (a/)} X r = (12) X r,
lowing theorem.
THEOREM. Angular velocities can be compounded by the Law
of Vector Addition.
EXERCISE
Prove the law of composition for angular velocities by means
of Equations (4).
172 MECHANICS
10. Moving Axes. Proof of the Theorem of 8. Let space
be moving as a rigid body with one point, 0, fixed. Let i, j, k
be three mutually orthogonal unit vectors drawn from and fixed
in space, and 7 be a second set of such vectors fixed in
let a, 0,
the body. The scheme of their direction cosines shall be the
following :
a j8 7 rj f
'l '2
(1)
Thus
a =
with similar expressions for /3, 7, where the direction cosines
are any functions of the time, t, continuous with their first (and
for later purposes their second or even third) derivatives, and
satisfying the familiar identities ;
cf. Appendix A. Observe that
187
=
(2)
aa = 0, etc.
07 + 7/3 = 0, etc.
(4) t = $a + 40 + fy.
A necessary and sufficient condition that P be at rest is, that
the projections of f on three non-complanar axes all vanish.
Hence, in particular, the condition that be at rest can be ex- P
pressed in the form :
(5) at = 0, fit
= 0, 7* = 0.
+ f ay =
(6) tfa+ f 07 =
=0
Let
(7) a = yp, b = ay, c = pa.
a = pj y
b = 7, c = aft.
These cannot all be unless the body is at rest, since the vanishing
of the above scalar products would mean that
pa = 0, ya = ;
(8)
= Xa, ij
= X6, f = Xc,
- oo < X < oo .
These points, and these only, are at rest. They form the instan-
taneous axis, and it remains to show that the* latter deserves
its name.
(10) (CO)
= C00! + ^0 + W^7.
Then (co) is collinear with the instantaneous axis, whose equa-
tions (8) can now be written in the form :
(11)
1 = JL = L.
CO^ COr, CO^
V{
= at = %aa + yap + f ay
= yt = %ya + rjyp + yy
174 MECHANICS
Hence
ff
= f
(co)
= co x i + coj + cosk,
+ a*,/? + co$
KINEMATICS IN TWO DIMENSIONS 175
EXERCISE
Treat the motion of the plane by analogous vector methods.
Let _
be the vector drawn from the fixed to the moving origin, and
let p,a- be unit vectors drawn along the positive axes of and rj.
Let r be the vector from the fixed origin to an arbitrary point P.
Then
r = f + p + r?<7.
* = f + fP + ^-
The instantaneous centre is given by setting t = 0.
p = e", a = C + D'.
culty.
and slides on S.
An intuitional proof can be given as follows. First of all, it
is clear from the very definition of L that S slides on S along L.
So it is necessary to prove only the tangency of the two surfaces.
Consider, first, the case that the moving axes have a fixed
0) r = f + 4/9 + [ 7.
Then
or
(3) v = v r + v,
where the terms on the right have the following meanings. The
vector,
,.. d drj _ df
^-dr + s'- -*"
, ,
<4> 1
(6) ve = () X r
or
(7)
(8)
(9)
= +
Hence
(11) v = v + v'.
Here,
FIG. 97 dx Q .
(12)
and v' is given by (9). The v' of (11) is, of course, not the V
of (8).
KINEMATICS IN TWO DIMENSIONS 179
EXERCISE
Denoting the components of v along the ( 17, f )-axes by ,
> v*y vl> show that the components of v along these axes are :
(13)
'dt
dy
+ ^~
4+
Here,
= 7*0.
being known :
(2) r = rc + r',
(3)
dt
of 14, 13 :
(4)
^(i> i/D fi) fixed in the moving space whose absolute vector
velocity is collinear with the vector (o>), or is :
(5) *,
= (<>).
(6)
Since
(7)
- = 0,
i
= i> i7i
= 61, fi
= ci
+ (fi
- c iK =
(1
- fli)
- (f i
- c^cu* =
Hence
(9)
and thus &, T^, ft is seen to be any point of the line through
(a,, bi, Cj) collinear with (<*>).
This line we define as L. These con-
ditions are sufficient as well as necessary.
KINEMATICS IN TWO DIMENSIONS 181
F: ti=F(t), 77!
= $(0, fi=*(0-
Its tangent vector at an arbitrary point is
k a + *?i** + #, 7 1 '
dt dt dt
tangent to C, provided it 5^ 0.
On the other hand, consider the point (| t 77^ J\) of F, for which ,
This last line is precisely the vector velocity of that point fixed
in S, which at the instant in question, t = t, coincides with
( x \> V\y z i)- This vector, t, let us call it, lies along L because of
(5), unless it be 0.
The other vector on the right of (11) is the vector (10); i.e..
a vector t 2 tangent to T at (, rj ly f,) or (x l9 y lt zj. Equation
(11) thus says that
ti
= t, + 1.
EXERCISE
Let a cylinder of revolution roll and slide on a second cylinder
which is fixed, the first cylinder always being tangent along an
element, and there being no slipping oblique to the element.
Choose the point (x i/ z ) in the axis of the moving cylinder,
, ,
(2)' T = -n
isthe period, where a is the amplitude, and where 7 is determined
by the phase.
In two dimensions, or with two degrees of freedom, an impor-
tant case of oscillatory motion about a fixed point is that in
which the projections of the moving point on two fixed axes
KINEMATICS IN TWO DIMENSIONS 183
x = a cos + 7) (nt
I
\u) 1
I
y = b cos (mi + e)
Example 1 n.: m=
Dynamically, this case can bo realized
approximately by the small oscillations of a spherical pendulum.
Let 7 = 0,
<p
= nt.
Then
mt + = e <f> + ,
= a cos <p
A = b cos e, B = b sin c.
cos = sin = x -
(6) <p
^,
<p
-^ ^ y.
(7) B 2
x2 + (Ax
- ay)
2 = a2 B2 ,
and this equation represents a central conic which does not reach
to infinity, i.e. an ellipse.
184 MECHANICS
The axes can be found by the methods of analytic geometry,
or computed directly by making the function
COS 2 ^ __ 2AB cos <p sin ^ + B* sin 2
^
x = a, y =b,
and continually touches all four sides, sometimes being a
diagonal, but, in general, an ellipse inscribed in the rectangle.
x = a cos nt
(9)
y = b cos (n + A/ + e)
cos 2* = or cos 2y .
where y denotes the angle from the axis of x to an axis of the conic. The length*
of the axes are found to be :
where
A 8 = a4 -f 2a*&* cos 2 + b*.
KINEMATICS IN TWO DIMENSIONS 185
is now + d,
I' =where
I d denotes the sag in the second string.
For small oscillations, the coordinates of the bob will evidently
be given approximately by Equations (3), and by suitably choos-
ing I and d, we can realize an arbitrary choice of m and n.
The Sand Tunnel.* If the bob of the pendulum is a tunnel
of small opening, filled with fine sand, the sand, as it issues from
the tunnel, will trace out a curve on the floor which shows ad-
mirably the whole phenomenon of the Lissajou's Curves. In
particular, if the second string is drawn as taut as is feasible,
so that d is small, the two periods will be nearly, but not quite,
<p nt:
Hence
(11) */
= - 6
I**
and the curve an arc of a parabola, passing through the vertices
is
A = b cos 6, B = b sin e.
f x = a cos nt
(16)
I
y = b cos [2n*+ ht + e]
Let <p
= at, 7 = 0. Then
= a cos p<p,
a<
(x = y
, cos
b(
6 (? + e).
27T
CO = M
Hence
X
- = /*
-,
.
Xg = ,,
them by P P
t, 2, Hence they must have at
) are all distinct.
circumference.
Turning now to the equations of the curve, let
= m= a.
<p nt,
n
Then
x = a cos <f>,
The curve cuts the line x = x' in the points for which
And now, since the angles 2kair lead to points on the circle which
are everywhere dense, the corresponding values of the cosine
factor are also everywhere dense between 1 and 1. +
It is of interest to study the multiple points of the curve. These
occur when
1
t) <f>'
= kw + ^, I ^0;
(19) -n * (1 + k a)ir.
(20) 77
= (1 +k Q a)7r,
then
a<p + f\
= a (<p + fc 7r) + i 7r.
Set
(21)
= <f> +k Q w.
x = a! cos
<22>
I =
where a! = a or a, and likewise &' = 6 or 6. Let
^ < oo.
x = a', y = &',
0' = kw + -a > 0,
where
/CTT
--> 0, I ^ ;
a.
or where
- far + - > 0, k ^ 0.
x a cos <p
z = c cos ($v +f )
a^xga, 6^2/^6, c ^ z ^ c,
qm 1 g
,
m < n, execute a Lissajou's motion. Thus a Black-
burn's Pendulum suspended in a moving elevator will have its
projection on a horizontal plane executing a Lissajou's motion,
whereas the vertical motion is not periodic at all.
ROTATION
1. Moments of Inertia. The moment of inertia of n particles,
w,-, with respect to an axis is defined as the sum :
/ =
(1) i><r<,
<=i
A =
(2)
(3) -0 = 5 m i Vi z <> E =
These definitions are extended in the usual way by the methods
of the calculus to continuous distributions.
In terms of the above six constants it is possible to express
the moment about an arbitrary axis through 0.
of inertia Let
the direction cosines of the axis be a, 0, y and let P (x, :
y, z)
be an arbitrary point in space. Then
r2 = p2
_ <^
or
r2 = x2 + tf + z2 - (ax + py + yz) 2
.
p
Since o
+P+ 7 = FlG
2
of 1,
-
(x
2
+ 2/
2
+ * 2 )(
2
+P+ 7
2
)
- (ox + fry + yz)*.
191
192 MECHANICS
Hence
- 2yazx -
Thus
/ =a 2 2
+ z> + 2 2
m 2
+ a* +
2
etc.
5) m;(^ ) 5J t-(*i )
or:
(5) Ax 2
+ By
2
+ Cz* - 2Dyz - 2Ezx - 2Fxy = 1.
meet the surface in the point (X, Y, Z), and let p be the length
of the segment of the axis included between the centre of the
(6)
2
p (^la
2
+ Bp + C7 2 - 2D/37
- 2#ya - 2Fa/3)
= 1.
P /
2 = / =
(7) 1,
^
and / is seen to be the square of the reciprocal of p. From this
property it appears that the Ellipsoid of Inertia is invariant of
(8) A(*
2
+ 2
2/ )
= 1,
non-degenerate) ellipsoid. If
coefficients A, B, all the ,
the centre of gravity, plus the total mass times the square of the dis-
I = J + MW.
EXERCISE
Show that the moment of inertia about any line, L, in space
is given by the formula :
*i
~ *> y\
- y> *i
- *>
194 MECHANICS
where #, ?/, 2 are the coordinates of the centre of gravity, and
x u Vn z \ are the coordinates of any point on L all referred to
the new axes.
(2) F(x, y, z)
= Ax 2 + Eif + Cz* + 2Dyz + 2Ezx + 2Fxy,
(3) *(*, P,s) = x* + y* + z*.
(4)
'
x* + ?/
2
+ z2 = a2 ,
or *(x, y, z)
= a2 .
(5) b\ + X*,
= 0, F 2 + X$ 2 = 0, t\ + X$3 = 0.
'
Ax + Fy + Ez + \x =
(6) Fx + By + Dz + \y =
Ex + Dy + Cz + \z =
But we know that the point (0, 0, f), f ^ 0, yields a maximum.
Hence = D 0, E= 0,
.
F(x, y, z)
= Ax* + 2Fxy + By* + Cz\
the coefficient of the term in xy does not vanish, it can be
If
(7) F(x, y, z)
= Ax* + By* + Cz\
Here, A, B, C may be any three numbers, positive, negative,
or 0, except that we have excluded as trivial the case that
all three vanish. The original equation (1) will obviously repre-
sent an ellipsoid if and only if the new coefficients A, B, C in (7)
are all positive. We have thus established the following theorem.
THEOREM. An arbitrary homogeneous quadratic function
F(x, y, z) can be reduced by a suitable rotation of the axes of coordi-
r 1
nates to a sum of squares. The new coefficients of x , y', z may
be any numbers, positive, negative, or 0.
EXERCISE
Show, by the method of mathematical induction and La-
grange's Multipliers, that an arbitrary homogeneous quadratic
function in n variables,
x{
= an x1 + + ai n x n
x'n = a n i#i + + a nn x n
such that, for any two corresponding points (x lt ,
x n ) and
(x( 9 ,
x'n ), the relation holds :
r '2
*l
_L
\
. . .
\
r '2
4. *n ~ r
*\
2 J_
\
. . .
\
r 2)
4. An
A = S an a nn ,
The function :
fl = F + \3>,
F' + X*',
an ~
_ 00 ~
~ ' ' '
a?
is equivalent to the condition :
A +\ F E
(8) As F B+ \ D
E D C +
ROTATION 197
arc seen to be the negatives of the coefficients A', B' t C', and so
the axes of the quadric are found. If the roots of the deter-
minant (8) are denoted by \ 19 X 2 X 3 the lengths
, ,
of the semi-
axes are
(1) r X A.
(2) ff = r X mv,
whore the vector drawn from the point to
r is
particle.
The moment of momentum of a system of particles with respect
to a point is defined as the vector
n
(3) ff = rk X m k vk
J) ,
1=1
* Cf. Appendix A.
f Contrary to tho usual notation of writing vectors in boldface, as a, x, i,
etc., or by parentheses, as (co), it seems here expedient to denote the vector moment
of momentum by a, the vector momentum by p, and the vector angular velocity
by o>.
198 MECHANICS
where rk is drawn from the point in question to nik, and v^ is tin
vector velocity of m,k.
In the case of a continuous distribution of matter the extensioi
of the definition is made in the usual way by definite integrals.
In Cartesian form a has the value, for a single particle :
J 1
y -
(4)
dx
'dt
m dy
dt
m dz
dt
dz di dx dz dx
dx k dz
(5)
dx
*L ( ^y. _ di - dz x
(&}
dt \ dt dt
y-^~v
(7)
=
(8)
ljt ^ mkTk X a *'
_
da = dv di
mrx __ + w xv^, .
^,
.
._*
Now,
-77
= v and v X v = 0.
at
Hence
do-
= mr Xy dv =
.
r * a >
dt dt
(1) I; m k (x k
^ - yk
d
ji*)
=
2
fc= 1
(Xk Yk
= Yk + Y
X
(2)
(3)
S-i>XF,
or:
= of the Applied Forces about 0).
^7 2) (Moments
(4) mk ak = F* + 2) Fw .
(5) m k Tk X a* = r X F* + r* X F*/,
fc
2)
equal to the vector moment of the applied forces, provided that the
internal forces between each pair of particles are equal and opposite
and in the line through the particles :
(1) M - .,
where v df/dt.
=
Equation (1) is merely the vector form of
Equations A), Chapter IV, 1. It can be derived by vector
Mi = 2)
= 1
(2)
A)
\
THE EQUATION OF MOMENT OF MOMENTUM :
- "
<
a
202 MECHANICS
and the vector cr remains constant. The line through collinear
with <T is called the invariable line with respect to 0, and a plane
perpendicular to the invariable plane with respect to 0.
it,
(2) a = a' + Mr X V,
(3) Mr X Q
v = (7
,
<T
O * 0.
CD r = r' +r
o r
;
f
Fia. 103 I V = v' + V ,
* x v*
or
Mr X v = r X Mv,
can be interpreted as the moment of momentum, relative to O,
of the total momentum, Mv, of the system, thought of as a mass,
M ,
concentrated at O and moving
1
with the velocity v.
Mr X v = f X Mv
is the moment of momentum, Mv, of the total mass, concentrated
at G and moving with the velocity of G, referred to 0.
204 MECHANICS
Let cr' denote the value of a referred to the point 0' as a fixed
point; i.e.
X vk .
Then
</ = m^i X vi + mk ri X v
5J ]
or
</ = v'r + Mr' X v .
(5) a = </ + Mr X v.
(I)'
S
And we have the Equation of Transformation, 8, (4) :
(2) a = *'r + Mf X v,
da _ d?
/ON
(6)
dt
~ dff'r
Ht
+M X Tt
On the other hand,
r = r' +f ,
(6) "
From it follows that
Mr X ^= f X
fc
F*.
respect to G:
EXERCISE
Show that a rigid body
is dynamically equivalent, in general,
(1) * = <j'
T + Mr X v + Mr' X v ,
and differentiate :
(3) 2 r" x F* = 2 ri x F*
t I
206 MECHANICS
From the Equation of Linear Momentum, 6, (1) follows that
Moreover,
V -f- V - .
-jj
For,
hence
v X v = v X v' +v X v = v X v',
and
v X v' + v' X v = 0.
g= 2r*XF*,
t
(6)
f=? ri>< F*
(7) f 'X =
for every value of t.
Let t = T be an arbitrary instant. Let 0' be a point which
describes a certain path,
(8) r - r ft r).
At the instant t = r,
/ dv *\
a2f
= / <A
(-5-),., (IF;,.;
Then Equation (7) is to hold for this vector r at the one instant
t = T.
~
I-JJ2 S (Moments about Inst. Centre).
208 MECHANICS
The only case, then, of motion in a plane, in which we may
permanently take moments about the instantaneous centre,
thought of as a point fixed in the moving body, is that in which
a circle rolls on an arbitrary curve, the centre of gravity being
at the centre of the circle; and the limiting case, namely, that
the point Q is permanently at rest. This last case corresponds
to the identical vanishing of dvjdt.
Moments about an Arbitrary Point. Consider now an arbi-
TO
= r (J, T),
dt dt 2
/-72/)
=
^77/2" 2} (Moments about Q).
EXERCISE
A billiard ball is struck full by the cue. Consider the motion
while there is slipping. Show that it is not possible to take
moments about the instantaneous centre.
Find the points of zero acceleration and verify the fact that
it is possible to take moments about them, explaining carefully
]
cf .
Chapter V, 8. Let the vector angular velocity be
denoted by w, and let a, p, 7 be a system of mutually perpen-
dicular unit vectors lying along Cartesian axes with the origin
at 0. When we wish these axes to be fixed, we shall use the
coordinates (x, y z) and replace QJ, j3, 7 by i, j, k.
y
In the general
case, the coordinates shall be 17, f ,
.
Let P be any point fixed in the body, and let r be the vector
drawn from to P :
a) r = $ + lift + f y.
The velocity of P,
(2) "
is expressed in terms of the vector co as follows (Chapter V, 9) :
(3) v = co X r.
In Cartesian form,
(4) v =
or
(5)
(6) a = r X rav,
y
(7) a =m r
Hence
(8)
210 MECHANICS
These formulas lead in turn to the following :
<T = m + f
2
) C0
- &CO,,
-
[(rj*
(9)
= m [
- rft + (f
= m [
- fcof - ftw, + (
2
+ i;
2
(10) or =-
dr =
dt
X F* = La + M/3 + Ny.
Thus we have :
= L,
dt
(2)
dt
dfft -LJ
~
C0 == xV.
_j." |
O"TJ WTJ (7A
at
On substituting for <T, o^,, a^ their values from (10), 12, the
/j
dp-
(3)
where
P = r =
When the axes of coordinates do not coincide with the prin-
cipal axes of inertia, Euler's Equations take the general form :
(4)1
dt dt dt
- (En,
- (C
- B) w^ = L,
dr
(3) A 2p2 +B 2
q
2
+C 2
r2 = I
= 0,
dt
and so
a = Apa + Bqf$ + Cry
is constant.
2
From Equations (2) and (3), two of the variables, as p and
#
2
can, in general, be determined in terms of the third,
,
and then,
on substituting in the third equation (1), a differential equation
for r alone is found. It is seen that t is expressed as an elliptic
Exercise. Let A = 3, B = 2, C = 1 ;
and let p, q, r all have
the initial value 1. Work out the value of t in terms of the in-
tegral.
ROTATION 213
(5)'
-* = ? = ,
p q r'
Thus (3) is satisfied. And (4) holds, too. Hence (2) is true.
S: 4 2
+ 5T; + 2
Cf 2 = 1.
1
= 1
= J* h '
P^l l
ment.
To prove the second statement; consider so much of the body
cone, (6), as lies in S. Let F be the curve on S which marks
the intersection of these two surfaces. Then F rolls on M with-
oujb slipping, and the curve of contact, C, can servo as a directrix
of the space centrode. For the body cone, (6), rolls without
slipping on the space cone, and the curves F, C are two curves
on these cones, which curves aro always tangent at the point M
of the instantaneous axis. The angular velocity, w, is propor-
tional to the distance Om ;
for
d$ dd
V = Sin COS <p rr + ,
SHI (p -r.
dt at
= ... <ty de
(1) q sin sin ip-j-
(it
+
,
cos tp-r
Ctt
'
dt dt
FIG. 105
with the point C in which the positive axis of f pierces the surface
of the unit sphere. As we look down on the sphere from above
this point, it is evident from the figure that
dO
= p sin <p + q cos <p
dt
V f
y
z
and the question is, to express the nine direction cosines in terms
of the Eulerian angles. This can be done conveniently by vector
methods, if we effect the displacement one step at a time. Let
i, j,
k be unit vectors along the original axes, and a, 7 unit /ft,
=
12 it cos k t
sin
J
=
2 Ji
k =
2 ij sin + kj cos 6.
13 = i2 cos + J2 sin
<p <p
j3
= i2 sin + cos v
<p j2
k,
= k,.
ROTATION 217
Zj
= cos cos <p cos ^ sin <p sin ^
m = l
cos 6 cos (f>
sin ^ + sin ^ cos ^
n!
= sin cos v?
dO de
P 3~T7* T =
~ = 0; hence r = v,
dt
d<p .
and at any later time, the result is the same. This can be seen
directly from the nature of the problem, since the motion of the
axis of f in the plane \f/ = is uniform, and hence the force which
the constraint exerts will be the same force relative to the body
at one instant as atany other instant.
It is easy to verify analytically the truth of the last statement.
EXERCISES
1. Show that, if no assumption regarding 6 is made, but ^ =
and r = v, then
d*e de
LT = AA + nCv cos vt -T:
-
4 . 4
sin vt -rz
at* at
d* e dd
M
TM = AA COS *
vt -T7T
1
at
~ n
Cv Sin vt -rr- <
at
dB
dO
37 =
c^
ce A
t
and , Q
9 = cA
-/ Y e
^
A
t
at CIJLV
This last equation is true, even when e varies with the time.
= 0.
(1)
C%
Hence r = v (constant).
The equation of energy here becomes :
(2) A (p 2 + 2
) + CV> = H- 2Mgh cos 6.
ROTATION 221
Bqn 2
On substituting for n lt n 2 nz
,
their values from 16 we have :
dt dO
p =
-
sin
<p-jr
...sin
= sin cos
dO
-=r
(4) q <p
at
(5)
2
sin e -- = - bi> cos
(6) 6 = C
(7) <p
= vt I cos 6 dt.
dt
2
- - -
sin 2 6 (~) = sin 2 e(a 2
(8) a cos 6) (J8 6i> cos 0) .
222 MECHANICS
The result is a differential equation for the single dependent vari-
(9) u = cos 9 :
(10)
= (1
- u*)(a
- au)
- CJ
- bvuY = f(u).
d* "" bvu
P
Hence \f/
is given by a quadrature after u has once been found
as a function of t.
Retrospect and Prospect. To sum up, then, we have reduced
the problem to the solution of Equation (10) for u as a function
of t.'
Equation (9) gives 0; Equation (11) gives \f/; and Equa-
tion (7) gives <p. We may concentrate, then, on the solution of
Equation (10).
(1) fM = (1
- u*)(a
- au)
- (0
- bmY
becomes positively infinite for u =+ oo. It is negative or
for u =+ 1, 1. Hence in general the graph will be as indi-
cated, or
</(M), u, <u<u 2 ]
fM = /(iO = o.
Moreover, 1 < u^ < u < 2 1, and
f(u) has one root, u' > 1. The roots
(a)'-
comes under the class discussed in Appendix B. In particular,
the solution is a function
(3) u = <*>(0
(4)
or
(5) <p(t +
Furthermore, if
=
(6) *!
then
And similarly, if
//w\ /, \
then
/*7'\
\t )
/*
^C*2 T)
\ ~ ^(^2
/j I
~T~
\
Ty.
(8) u = u lt u = u2 .
III. < 0.
FIG. 109
CASE I. Since
bvu
eft 1 - u2
is positivewhen u has its greatest value, u 2j d^/dt will remain
positive, and so ^ will steadily increase with t. Let ^ = when
t = 0. As increases to iT,
t will increase to \!/
224 MECHANICS
where u =
^(0, Equation (3). When t T, \p will have in-
=
creased by ^, and one complete arch of the curve C will have
been described. The arch is symmetric in the plane ^ = -J-^.
The rest of C is obtained by rotating this arch about the polar
axis of S through angles that are multipla of ^.
|8
- bvu 2 = 0.
and so \[/ steadily increases. The curve C has cusps on the upper
parallel of latitude.
advancing ^, as indicated.
Special Cases. There is still a variety of special cases to be
discussed, one of which is that in which f(u) has equal roots
lying within the interval :
1 < MI
= tig < 1.
Since
/(I)
in all cases, and since
1
there must be a third root u ^ 1. Thus u is a double root and
f(u) = (w
- u,
where
x(u) < 0,
- 1 < u < 1.
u = wt .
EXERCISE
Treat the case of a top on a smooth table. Assume that the
peg is a surface of revolution. The distance, then, from the
centre of gravity to the vertical through the point of contact
with the table will be a function of the angle of inclination of
the axis.
Assume axes fixed in the body with the origin at the centre
of gravity.
Write down i) the equation of energy; ii) the equation that
says that the vertical component of or is constant.
From this point on the procedure is precisely as before, and
the result again a differential equation of the type treated in
is
Begin with the special case that the peg is a point. Having
studied this case in detail, proceed to the general case and study
it in detail, also. Then derive the special case as a particular
case under the general case.
v '
__ ds ds
.
Since
tan V = , . .
-, or V= tan" 1 -77-; -,
d\f/ sin B $' sin 0'
it follows that
(3) jc
= WB" - 0' iH sin - (1 + /2
) V cos 0.
* The point O need not be the centre of gravity in the following treatment.
It may be any point fixed in the axis of material symmetry.
ROTATION 227
The cone
lies to the right of the observer, as he travels along C.
U Ax
-
1 f
be written in the form :
ntt
(5) K =
* These results are inserted for completeness. They will not be used in what
follows, and the student may pass on without studying them. They are chiefly
of interest to the student of Differential Geometry.
228 MECHANICS
If K is known, or given, as a function of s, then Equation (6)
determines as a function of s, and is then found by a quad- \l/
rature :
(8)
i j k
(
(10) n = a X t x y z
(11)
hence
= yz"
-
(12) zx" - xz"
KZ
'
= xy
_ yX ".
Since I
K\ = 1 1' I and
formula (9) follows at once from (12) and (13). Moreover, from
(12) it follows that
x y z
d
Ts
( v *=
l
T,
* For the
proof of this formula cf. the Author's paper cited above.
ROTATION 229
(1)
us, namely that in which there is only the force F (and the reac-
tion at 0) we have : N
= 0, and so r = a constant. *>,
(2)
AKV* + Crv = Q,
*Cf. the Author's paper "On the Gyroscope" cited above, p. 240.
230 MECHANICS
Proof. Let the unit vector from to P be denoted by a (it
isthe vector 7 of the coordinate system) ;
let t be a unit vector
along the positive tangent to C at P; and let n be a unit
vector normal to a and t and so oriented
' a
with regard to them as ft is with regard
\p X to 7 and a. These are principal axes of
t \ inertia, and the moments of inertia about
them are :
L = A In = A, Ia = C.
FIG. Ill
o>t
= 0, wn = I),
a = In C0 n n + It CO* t + Ia C0 a a.
Hence
(3)
do- dv dn da
It is clear that
da
(4) vt.
dt
/^x dn =
(5) KVt
-j-r
Hence, finally,
d<r
(6)
= Av-n + (Aw* + C)t + C
Let the vector M
which represents the resultant moment of
all the applied forces about be written in the form :
M = M n n + M t t +M a a.
Since
da -.
ROTATION 231
we have :
(7) Av^-
us
= M n, A Kv* + Crv = M t, C% =
dt
M a.
F=-Qn+Tt,
we see that M n = T, M t
= Q, and thus Equations (2) are
established. Equation (1) is the third of Equations (7).
We have thus obtained Euler's Dynamical Equations in the
form:
dv
Av^- = T
ds
shows that the component r of the angular velocity & about the
axis of the gyroscope varies exactly as it would if the axis were
permanently at rest and the same couple N relative to the axis
acted.
The second equation,
A) AKV* + Crv = Q,
expresses the sole relation which holds between the four variables
K, v, r, and Q. In the applications, however, r is constant, r = v,
and so the equation
A') AKV* + Cw = Q
expresses the sole relation between K, v, and Q.
232 MECHANICS
The Case F = 0. Let us begin with the case that F vanishes,
but the axis is not at rest. Here, Q = 0, T = 0. Equation A)
gives
i) AKV + Cr = 0,
Av
p==
'Cr'
Cv
A __ n
FIG. 112 ds~ '
cot a = |
K |,
or tan a = p,
Am* + Crv = Q.
The left-hand side is the sum of two terms. The second term
expresses the force,
Q2 = Crv,
ROTATION 233
a = xi yj + zk.
Then
v = xi + y] + zk = vt,
where x' = dx/ds t
x = dx/dt, etc. Furthermore,
n = a X t = (yz
f
- zy ) i + (zz' - xz ) j + (**/' -
f f
yx')k.
The acceleration, (a), of P in space is, of course :
() = zi + j + ^k.
Now, the component of the acceleration along the normal n
to the plane of a and t is n-(a), which can be written in the
form :
x y z
f
x' y z'
x y z
Since x = vx', it follows that
point P, in which the axis of the gyroscope meets the unit sphere
about 0, moves like a particle of mass = A constrained to m
lie on the sphere and carrying a charge of electricity, e. The
forces that act on m are supplied by the electromagnetic force
of the field, Q 2 = Cw, acting on e and a force F acting on m, y
* The
idea of using the above electro-magnetic field to obtain #2 was suggested
to me by my colleague, Professor Kemble, to whom I had just communicated the
results of the text, down to this point. (Note of Jan. 23, 1933.)
ROTATION 235
EXERCISES
1. Suppose the axle P of the gyroscope is caused to move in
a smooth slot in the form of a meridian circle, which is made to
rotate in any manner. The force F will then be normal to the
meridian, or tangent to the parallel of latitude. Show that
d2 6 .
equations :
=
Av%
ds
T,'
AKV* + Cvv = Q,
K =
where the first term on the right is due to the damping of the
water; the second, to the righting moment produced by the
buoyancy ; and the third, to the force exerted by the trunnions.
The frame may be thought of as rotating about the point, 0,
dt
where the first term on the right is due to the brake and other
damping, and the second, to gravity, since the frame is so con-
structed that its centre of gravity is appreciably below O.
On combining these five equations and neglecting. A + A' in
These are the equations which govern the motion. They are
discussed at length in Klein-Sommerfcld, I.e.
CD .
rf2
2
dt
where
f Ma
2
.
thus gives :
/- = nMga sin
u-6
'if- -
Hence
(3)
= const.
dt* 5 dt
(4)
^y.
dt 2 5 dt
Hence
2a
(5)
2a
(6)
Vv = V sin ^ = -
+ au(.
2\dt
^
where
A' =- B' =- fB.
Equations (1) now take on the following form. For abbrevi-
ation let
A',
dt
Then
du
dt
(8)
dv
Hence
du dv ^
v -7T u -77 = 0,y
dt dt
and consequently
av,
av av
(9) cos sn = ft
(10)
au( | t-o
= 0, aco, \
t -o = c, wf |
t-o y,
Mw =
(12)
- -
~ Y
Y>
dt*
where X, Y
are the components of the unknown reaction at P.
Next, taking moments about the centre of gravity, we find :
(13)
UOi)
___. f UWf)
i ___. _ ax -w*
dt dt
dt dt
ROTATION 241
Finally,
Tr
y. - - - a*, - A
(14)
But the left-hand side of this equation vanishes because the first
equation (14) is an identity in t. Hence = 0. Similar con-
X
siderations show that Y = 0.
Md x
-;- X cos 6 Y sin 6 + Mg sin 6
(1)
= Xsin0 + Ycos0
Moreover,
=- da, 7 = 0.
(2)
dt
ROTATION 243
Since
= 0, co,,
= <p, o>
and A = C = a2 ,
# = pfa 2
,
we have :
Hence
or
(3)
=- Beta. + 5^j3 +
forces :
at r2 =-0.
Thus
or, finally :
(5)
dx dy =
= v cos 0, v sin ^,
~dt dt
244 MECHANICS
where
and so
/n\ dx = a--
d<p n dy
-~ = d<p .
(6)
v ' -TT cos 0, a- sin 0.
dt dt dt dt
3/\
= 9 = \t + M
(7)
ft
X, .
Ma - = i Ma
~ + Mg sin e cos 0,
or
(8)
where
Hence
(9)
^ = *(sin-slnM) + *o,
and
^ --
v?
= fc
,
(cos M
- M+ cos 0)
. / . fc sin M\
+
.
v?o-
^5 Jt
EXERCISES
The student should first, without reference to the book, repro-
duce the treatment just given in the text, arranging in his mind
the procedure: i) figure, forces, coordinates; ii) motion of the
centre of gravity; Hi) moments about the centre of gravity;
iv) conditions of constraint v) the solution of the equations.
;
taken into account. Begin with the case that the bottom of the
frame is smooth and its centre of mass is at the centre of the disc.
4. Study the motion of the centre of gravity of the disc treated
in the text, by means of the explicit solution of x, y in terms of t.
gravity.
The Principle of Work and Energy frequently gives a useful
integral of the equations of motion.
If the right-hand side of the Moment Equation is a vector
lying in a fixed plane, the component of a normal to this plane is
constant, and thus an integral of the equations of motion is
obtained.
Sometimes there are conditions which are expressed by equa-
tionsbetween time-derivatives of the first order, t = t, but which
cannot be expressed by equations between the coordinates only.
246 MECHANICS
The first step in solving a problem is to draw the figure, mark
the forces, and pass in review each of the items just mentioned ;
(1) W = Fl,
F=f(x)
be the force, where f(x) is a continuous function. Divide the
interval into n parts by the points X Q = a, x l9 ,
x n -\, x n = b,
Fk and consider the fc-th sub-interval :
3*-i ^
-
x ^
_,
xk
.
Az* = xk - _!.
FlG 116 ,
ii) the work for shall lie between the work cor-
any interval
responding to the maximum
value of the force in that interval,
and the work corresponding to the minimum force :
g AW ^ k
248
WORK AND ENERGY 249
where
Fi F Fi'
FZ =/(*;'), **-
i>
Hm = f
J
Hence the requirements, i.e. physical postulates i) and ii) are
sufficient to determine the definition of the work in this case * :
(2)
W=Jf(x)dx.
a
(3)
W=Jf(x)dx.
a
Hence
(O
*
f- ,.
Strictly speaking, we have shown that (2) is a necessary condition for the
definition of work according to the postulates i) and ii) . It is seen at once, however,
that conversely Equation (2) affords a sufficient condition, also.
250 MECHANICS
EXERCISES
1. Show
that the work done in stretching an elastic string is
&
(5)
W=ff(x)dx.
X
j*a
must lay down our definition
r^ - -~- i
of work so that the work done
\ii \
(6) W = Fl cos ^.
x2 x l
= I cos T, 2/ 2 2/i
= Z sin r.
Hence
(7) TF = X(x z arO H
Jb-l
FIG. 118
where Zfc denotes the length of the
chord, gives us approximately what we should wish to understand
by the work, in view of our physical feeling for this quantity.
The limit of this sum, when the longest h approaches 0, shall be
defined as the work, or
(8) W
Since
~^-
= 1,'
As*
* Cf. the author's Advanced Calculus, p. 217. It is imperative that the student
learn thoroughly what is meant by a line integral.
252 MECHANICS
A second formula for the work is obtained by means of (7) :
< i
(10) W= Ax COST +
J
7 SUITES =
JC(x^
^ as
+ Y-
t
or
or
(', 6'. c')
Cxdx+Ydy + Zdz.
(a.b.c)
W= Zdz = = -
jXdx + Ydy + mg(z l z ).
jmgdz
C ZQ
Hence the work done is equal to the product of the force by the
difference in level (taken algebraically), and depends only on
the initial and final points, but not on the path joining them.
EXERCISES
1. A well is pumped out by a force pump which delivers the
water at the mouth of a pipe which is fixed. Show that the
work done equal to the weight of the water initially in the well,
is
X = 2x 3y + 4z 5, Y= z x + 8, Z = x +y+ + z l2.
WORK AND ENERGY 253
Find the work done when the particle describes the arc of the
helix
x = cos 0, y = sin 0, z = 70,
C: x= /(X), y = *(X),
= f (X), X S X ^ Xi,
(1) F = Xi +Y + j Zk.
The example serves to illustrate the general idea of a field
of force. We may have an electro-magnetic field, as when a
straight wire carries a current. the north pole, P, of a magnet
If
is brought into the neighborhood of the wire, it will be acted on
Z-0,
where (r, 0, z) are the cylindrical coordinates of P, and C is a
positive or negative constant. Thus in vector form
and
254 MECHANICS
Force Function. It may happen that there is a function
(7) V = i +' + k -
Then Vu is defined as :
Hence
(9) F = Vu.
(10)
=
where r is the distance from the given fixed particle to the variable
particle, and X is a positive constant.
-
(n) 2T k
>
(12) 11 = CO.
We may also write :
(13)/
v u = C tan~ l -; 9
X
but this formula is treacherous, since only certain values of the
X = |H = C-=g- Y= -
C^jhi, Z= 0.
(14)'
dx x + v2
2/
2' f*
?/ x + y
2 2
Work. When
a particle describes an arbitrary path in a field
of force, the work done on the
particle by the field is given by
Equation (12) of 2. If there is a force function, this formula
becomes :
i i
&u 7 . du , \
(17) W = u(x, y, z)
- u(a, 6, c).
(*.*.
dx dy dz
(20) * =- u + C.
In case, then, a potential <p exists,
256 MECHANICS
EXERCISES
1. Show that the field of force defined by the vector (3) is
not conservative. But R
be any region of space such that an
if
W = Km(- -
Second Law :
(i) '
dzd^z\ _ ydx
~ -dz
} m ~
r ^ ,
ydy +
\dt~dt 2 ~dt~dt
2
dt ~dfi) dt dt dt
v *'
~2dt
where
=
Hence
v2
W+ ~dP
+ ~dt
2
'
md 2 Y dx ,
y, dy t ^ dz
2 dt dt dt dt
each side between any two limits, < and t lt the results must
tally :
f*. Vdt
dt =vt (
J
o
2dt J
'o
\
taken over the path of the particle 2, (13). But this is pre-
;
cisely the work done on the particle by the force that acts. Hence
The quantity
mv
2
d xk v
,.^
(4)
mt-^r-Xt, *-- = ,
we infer that
^ (v
2( X
xk
dx
+^
.
v dy k ,
7 dz k \
89 r.-'.-
t
(7) 1\ -T = U,
- U
The potential energy, <l>,
is defined as :
(8) $ =- U + const.
(9) 7
1
! + * = T + *o-
t
(10) E = T + *.
We have, then :
(11) E, = Em
or the total energy remains constant. This is the Law of the Con-
servation of Energy in its most general form for a system of
particles.
(1) F ty + Fn = 0,
WORK AND ENERGY 259
and lie along the line joining the particles, and furthermore the
(2) rl
= (Xi
- xtf + ( yi
- 2/y)
2
+ (Zi
- ztf
is independent of the time, or
/ + Y^ dyi + Z/ dzi
C( Y {
-L- v ^Mi -u 7 ^A ,//
Since
Xu + Xn = 0, Ya + Y,-t
= 0, Za + Za = 0,
~~ dy\ dz
'
i
+v (dyi ~ ~ (dz { " ~
ij Yii
\dt
/(
This last integral vanishes. For, the force F i; is collinear with
the line segment connecting w -
t with m
jf or :
(i) T =
Hence
(3)
i-Tr+TT
Let v denote the velocity of the centre of gravity, G ;
and lot
Moreover,
I = 7 + Mh\
where 7 is the moment of inertia about a parallel axis through
G. Hence
(4) r
WORK AND ENERGY 261
of like mass, moving with the velocity of the centre of gravity, and the
kinetic energy of rotation about an axis through the centre of gravity,
= 770^ fa?,/
~~ w
fw
Vt
Hence
(6) T = i
Km cos
FIG. 120
(2) W= I Fvcostdt;
(3)
=
/Fvttt,
?r
= ds ds, x,.
-TT at = -=- aA.
j.
v at
WORK AND ENERGY 263
the plane.
Here, C is either the straight line or the
F = cos
ds dd
and
'
cos {(! s )
- <* (0i
- 0o)l-
_ ds d8
v a
~Tt dt
EXERCISES
1. Check the result in Example 2 by determining the motion
of the ball and computing the change in kinetic energy.
2. A train is running at the rate of 40 m. an h. The baggage
car is empty, and the small son of the baggage master is disport-
264 MECHANICS
3.
,
A
**!
rope
^~^~"^
^
is
B
frozen to the deck of a ship.
I
The free
haaled over a smooth pulley at P.
,
5. Show that the internal work due to the rope in an At wood 's
machine is nil. Would this be the case if the rope stretched?
6. A number of rigid bodies are connected by inextensible
cords that can wind and unwind on them in any manner without
slipping. Show that the sum of the works done by the; cords
on the system and the system on the cords is nil. First, extend
the definition of work so as to include the case of the work done on
the system by the part of a cord which is in contact with a body.
where s denotes the distance the escalator has moved since the
man came aboard
WORK AND ENERGY 265
(2) W
And
(3) I = c* t ,
where I is the distance the escalator has moved while the man
is running up.
On the other hand, consider the motion of the centre of grav-
ity of the man. Let the axis of x be taken up the plane. Then
M --f = S Mg sin a,
(4) Mu^ Mu = I S dt sn a,
where u dx/dt.
It follows, then, from (2) and (4) that
If the man
steps off with the same velocity with which he
stepped on, u^ =
w then, with the help of (3),
,
Now
h = I sin a
is by which the man would have been raised
the vertical distance
in tho time he was on the escalator if he had not run, but stood
still. Hence, finally,
(7) W = Mgh.
It makes no difference, then, whether the man runs fast or
slowly, up or down. The one thing that counts is how long he
is on the escalator. Thus when small boys play on the escalator,
running up and down, the work the escalator does increases in
266 MECHANICS
proportion to the time they are on it, provided they arrive and
leave with the same velocity.
--
---^[
^p^f
-
regarded as a point fixed in the one body,
must be the same as the vector velocity of the
point of contact, regarded as a point fixed in
>^
Vi' v2 the other body.
The works done by the two forces Fj, F on
2
*i i
W 1
= CF.V, cos ft dt, W 2
= (*F 2 v 2 cos ft dt.
to
But F = F
l 2, i\
= v2, ft + ft
= TT. Hence
W, + W 2
= 0.
v l cos ft + vz cos ft = 0,
Since furthermore F = F we l 2)
have :
W + W 2 = 0.
1
Fio. 126
m
w ^i _
2
2
-fl^o
2
2
'
(2) W=
Since there is no rotation, v = v, ^ = ^,
and
(3) w= CFvcos$dt.
A dp .
"" n dq .
"
~ dr ~ T ,
,, ,
,r
Hence
a (b -
X c)
=
Moreover,
w X r =- (r X o>).
Hence
F-(w Xr) = -(r X F).
From (5) it follows, then, that for a single force, the change
in kinetic energy is equal to the work done. For the case of
n forces the proof is now obvious. The extension to the case of
body forces and forces spread out continuously over surfaces or
along curves, presents no difficulty.
Remark. We have shown incidentally that the work done
on a rigid body with one point fixed is
h
o>
J t,
where
M = La + Mp + Nj
is the resultant couple.
The General Case. Consider first a single force, F. The work
it does is
W = /Fvd*.
Here,
V = V + V',
where v is the velocity of the centre of gravity and v' is the ve-
locity of the point Q relative to the centre of gravity, as it flashes
through P. Hence
W=
WORK AND ENERGY 269
Mv*
EXERCISES
1. A ball is placed on a rough fixed sphere of the same size
and slightly displaced near the highest point. Find where it
will leave the sphere. Let p, have any value.
2. A weightless tube can turn freely about one end. smooth A
rod is inserted in the tube and the system is released from rest
with the tube horizontal. How fast will it be turning when it
is vertical ?
IMPACT
1. Impact of Particles. Let two particles, of masses m l
and
W 2, be moving in the same straight line with velocities Ui and
w2 ,
and let them impinge on each other. To find their velocities
after the impact.
Isolate the system consisting of the two particles. Then no
j* 1 w2 external forces act, and so the
/$|Q xgfr
(1) m^ m^! +m 2 u2 w 2 w2 = 0.
As yet, nothing has been said about the elasticity of the par-
ticles. The extreme cases are: perfect elasticity (like two bil-
u(
= u 2,
u = miUl m "*" * u*
(2)
m +m
l 2
T T
<-T / t=T /
(4) mu l l
= I Rdt.
J
i-o =
J / e/
(5) P = CR dt.
o
Hence
= P
(6)
m U 2
= P.
The second stage of the impact now begins, as the balls are
kicked apart by their mutual pressures. On integrating the
V
equations (3) between the limits T and 7 we have ,
:
T'
(7) m^ fftdt',
CR'dt'.
tunate, since P
is not of the nature of a force, which is a push or a pull, but rather
is expressed by a change of momentum. Moreover, the dimensions of impact are
ML/T, not ML/T*.
272 MECHANICS
ticity. Hence
7" T
(10) P' = /V dt f
= e CR dt = eP.
(U)
{ ^-Ilf/ = df
The four equations, (6) and (11), contain the solution of the
problem. Between them, U
and P can be eliminated, and the
resulting equations can then be solved for u[, u2 . The result is :
Uf\*
m 4-
+ m, vn..
l
:
(12)'
e) m u + (ra
1 1 2
U2
U2 = U 2 .
mass m l after the impact will be given by the first equation (13),
but they suggest it. The proof is given by means of the first
of the equations (6) and (11), resulting as they do respectively
from the first of the equations (4) and (7), combined with (10);
U having here the known value u 2 .
If, in particular, u 2
= 0, we have :
(14) u{
= -,.
Perfect Elasticity, e
*= 1.
Equation (14) becomes in this case
u[
= -MI, and the ball recedes with the same velocity as that
with which it impinged.
If the masses are equal, m =w
l 2, Equations (12) become :
IMPACT 273
P' = eP,
as the physical postulate governing impact.
EXERCISES
1. A ball of 6 Ibs. mass, moving at the rate of 10 m. an h.
overtakes a ball of 4 Ibs. mass moving at the rate of 5 m. an h.
per cent heavier than the other one, in order to attain complete
rest for the latter. What per cent larger should its radius be ?
8. If, in the last question, the pendulum bobs are of glass.
e = ^f ,
find the ratio of their diameters.
(15)
where
X = # cos 6, F = 72 sin c,
(16) rn^X, *, F.
^
,.,,_>.
(17)
=- /-Yd*, m,^
1
=
*-o J dl * =o J/Yrf*.
And now we denote the first impulse by P, and lay down the
postulate that the second impulse is :
(18)
m^U m l
u l
= P m F t m^^
[
(19) 1
= P
which hold for the first epoch of the impact, the equations for
following :
m } u\ m,U= eP { m* v\ m V=}
1.
4 m U =
2 eP I
m^ m F=2 0.
(22) v[
= v l9 v'2 = v2 .
276 MECHANICS
The components of the velocity along the line of centres are
u\ = - *--
mi
(23)
Mj
=
m l
r
~~2 ~ _
" ~~>\ ,
'
EXERCISES
1. A smooth ball travelling south-east strikes an equal ball
- = Y
*'
dt*
7(0'
- 0) = Ax - x) Ydt - - y)Xdt.
f(y
Concerning F we
assume that the vector changes con-
will
tinuously in
magnitude and direction during the interval of
time in question, and that the point of application, (x, T/), also
moves continuously, remaining near a fixed point (a, b) through-
out the interval. Let
(3) x = a + , y = 6 + 17.
(5) /(' - 0) = (a
- *) Q - (b
- y) P
T T
+ CtYdt - Cr,Xdt.
M(u'-u)=P,
(6) '
7(12'
- $2)
= (a
- x) Q - (b
- y) P,
(7) 7(Q
; - fl)
= M(a - x)(v'
- v)
- M(b - y)(u'
- u)
or
(8) fc
2
(12'
- Q) = (a
- x)(v' -v)- (b- y)(u'
- u).
An Example. A
rotating about one end, and it strikes
rod is
O k
{Let '
be
Let
h, and let I be the length of the rod.
Since
M/ 2
u = 0, fi' = 0, ^ = 0, 12' = 0, / =
^-,
we have :
Hence
(9)
//* ft
f(x) <p(x) dx
g Y.
Hence
C^Ydt =
'
(10)
f*Ydt,
where
'
is the value of at a suitable point, t = ',
in the interval
7
g g 7
Now, by hypothesis,
.
T? approach uniformly, and
i.e. the largest numerical value that either has in the interval
^ t g T approaches 0; and furthermore, also by hypothesis,
the integral on the right approaches a limit, Q. Hence the inte-
gral on the left approaches 0.
If therange of the angle of F does not exceed 90, the axes of
coordinates can be so chosen that neither nor Y changes sign X
7
in the interval ^ t ^ 7 and then it can be shown as above ,
x' = ax + by
(11)
y
r
= ex + dy
where
a = cos 7, b = sin 7,
FIG. 134
c = sin 7, d = cos 7.
r = of + 6,
0000
Hence
r r T T T
The integral on the left, and the last two integrals on the right,
with 7 as has been shown above. We
7
will show that this is true also of the other integrals, and hence
in particular of the last integral in (5). To do this, write down
f
= p cos ^, = p sin
77 \l/.
extremities, on an arbitrary line, is the same for 'both lines, and ends by declaring
the coordinates of a point as the projections on the axes of the vector whose
initial point is the origin and whose terminal point is the point in question. With
that definition, Equations (12) and (13) are merely particular cases of Equations
(11), since both sets of equations express the projections of a vector on the coordinate
IMPACT 281
_ 2wt TT
_ 2wt
Then
r T
P-fx* Ffa*-0,
rpy
/(*r-
to demand that Y^ 0? That this is not enough, is shown by
modifying the above example as follows. Let
X = F cos ?, Y = 0.
The integral then has half the value it had before; hence, etc.
EXERCISES
1. A uniform rod at rest is struck a blow at one end, at right
angles to the rod. About what point will it begin to rotate?
the box, before it reaches the curb, comes to a bare spot, M = 1>
what is the condition that it should not tip?
282 MECHANICS
4. If the box tips, find whether it will slide, or rotate about
a fixed line.
6. Show
that greater braking power is available when the
brakes are applied to the wheels of the forward truck of a rail-
road car.
no
where
^^^^
F=pR
Fid. 136
V = 0,
We have, then :
Q = M(- t>
), vl = e(- V Q),
^
"
__ _ _ __
the slope of the trajectory at the end of the impact
u + (1
e(
+
VQ)
e) M (-t,
= ( VQ)/U Q is the numerical value of the slope before
)
~
__
1 +
e\
(1 +
;
e)/i\
i.e.
where X
the impact.
As the ball has a drop due to the cut, X will be considerably
larger numerically than the slope in the part of the trajectory
just preceding the last ten feet or so before touching the ground.
It might conceivably have a value as great as The value of .
Thus
& = .15,
*i
u,
= w [l + (1 + <0/*A]
= 1.09t* .
EXERCISES
1. A billiard ball, rotating about a horizontal axis, falls on
rough, M
= i-
6. A
rigid lamina, at rest, is struck a blow at a point 0. Find
the point about which it will begin to rotate.
begin to rotate ?
CHAPTER IX
RELATIVE MOTION AND MOVING AXES
1. Relative Velocity. It is sometimes convenient to refer the
motion of a system to moving axes. Let O be a point fixed
in space. Let 0' be a point moving in any manner, like the
centre of gravity of a material body, or the centre of geometric
symmetry of a body whose centre of gravity is not at 0'; it is
a point whose motion is known, or on which we wish particularly
to focus our attention. Finally, let P be any point of the system
whose motion we are studying. Then
(1) r = r + r',
dr " dr
_ ^ di^
dt dt dt'
or
(2) v = v + v'.
, i
FlG 137 , i
-
(4) r = a + iff + f y
and
+ r/4 + f 7-
285
286 MECHANICS
This equation represents an analysis of the velocity
(6) v =
(7) v = vr +v e,
where
(%} v = 4- #4-^
isthe relative velocity of P with respect to the moving axes ; i.e. the
absolute velocity which P would have if the ( r;, f )-axes were ,
(9)
= f(t), ??
= <p(t), f = \l/(i).
Secondly,
(10) Ve = a + rift + f7
(11) (w)
= pa + qp + ry,
where
Then
FIG. 138
(12) ve = () X r,
or
(13) + (&
- f p) j8 + (77??
- y.
(14)
RELATIVE MOTION AND MOVING AXES 287
Hence
(16) a = + ,
or
(17) a = a + a'.
__ dv* *
__
(20)
= dVrt
~
a$ jT-
+ pity qv$
+f + 77/3 + f y.
Thus
o _
|
y* dt
!
c^ l
<ft e&y
+ <* + r/j3 + f 7.
(23)
is
ae==
^ +
the acceleration d'entrainementj or the SchleppbesMeunigung,
77
d^
+ f W~
(24) a = r/3
- qy, $ = py - ra, y = qa - p/3,
we have :
(25)
-J-}
+ rp - qy.
where
(co)
= pa + q@ + ry.
Hence
/
/oc\
(26) * - /\dQ
r
- '
dr\ a
+ f.dr - >.dp\
f ' + "
dp -
- w2 ( + i?/3 + r 7) + (p$ + qr> + rf)(pa + g/8 + ry),
or
(27) ae = (0 X r - 2
r + (()r) (),
where
This vector (w') is the velocity relative to the fixed axes (, 77, f),
with which the terminal point of (co) is moving when the initial
point is at 0' it is the relative angular acceleration, referred to the
;
a' =
dtdt^ dtdt^ dt dt
j
(30) a = (co) X vr ,
or:
hand side of (10) comes to coincide with the right-hand side of (29).
With the aid of (12), this vector can be written in the form of the
right-hand side of (30), and this completes the proof.
To sum up, then : From (17),
(31) a = a + a',
(32) a = ar + 2a t +a e,
(1) ma = F,
it follows that
(2) ma + ma = F,
a = ar + 2a, + a*.
(3) ma = F
r ma Q 2ma ma e.
On substituting for a* its value from (30) and for a e its value
from (26), a system of differential equations is found for deter-
mining , 77, f :
rf \dt dt dt
(4)
a,, a{ .
RELATIVE MOTION AND MOVING AXES 291
EXERCISE
Obtain the dynamical equations in explicit form from La-
grange's Equations, Chapter X.Observe that
f 6? +& fp + tfop)
2
where
_
~ V '
7 -
__
* + + "I -5-
Vy 3
-ft-
*
-fa
reduces to :
(1) a.=-*r+(().r)(),
and is Kinematically, it is, of course, the
easily interpreted.
centripetal acceleration ; geometrically, it is a vector drawn from
the point P
toward the axis and of length o> 2 p, where p is the
distance from P
to the axis.
Newton's Law takes the form :
ma = F,
(2) a = + 2q
- 2r - + pfa + 9, + rf )
292 MECHANICS
Axis of f, the Axis of Rotation. In this case, p = q = 0,
r = w, and the equations reduce to the following :
(3)
p = co cos X, q
= 0,
r = w sin X,
2rr
co = = .000727.
24 60 -
60 FIG. 140
ma + ma = F.
Let P be the point of support, and let (f, 1?, f ) be the coordinates
of the pendulum ; Z, its length,
First, compute F :
F = G + N,
where
- au ,
w.c>u
is the force due to gravity, or the attraction of the earth ;
and
N = - j^a-Jtffl- j-#7
a = co
2
p( sin X + 7 cos X).
Finally, a
given by the formulas (2) of
is 5. For, although
these were written down for the particular case a = 0, they apply
generally, where a is arbitrary, provided the vector angular
2
Next, suppress those terms which contain co as a factor, or are
of the corresponding order of small quantities. Thus
Finally,
?+ T? + f
2 =
= -I + r?
+ terms of higher order.
A)
dtdt*~^dtdt*
2~ Z 2
or
= +C
(2)
fJ 'r f
hence C = and
dB
= "'
di
It follows, then, that
= w't.
motion.
The General Case.Returning now to the general case, let
the motion be referred to amoving plane through (the point
of equilibrium of the pendulum), perpendicular to the f-axis,
and rotating with constant angular velocity a/ about 0. Then
(3) <f>
= B - u't
(4)
r-l-C,
and this is the equation of areas in its usual form.
or
(5)
^ + r^ + 2'C + rV = - |r 2
+ .
296 MECHANICS
On suppressing the term r
2 '2
because of its smallness, we find :
+--!-+*
But this is precisely the equation of energy corresponding to an
EXERCISE
Obtain the equations of motion A), directly from Lagrange's
Equations, Chapter X.
CHAPTER X
LAGRANGE'S EQUATIONS AND VIRTUAL VELOCITIES
INTRODUCTION
In the preceding chapters, the treatment has been based on
Newton's Second Law of Motion. Work and Energy have
entered as derived concepts. It is true that certain general
theorems have been established, whereby some of the forces of
constraint have been eliminated, like the theorem relating to
the motion of the centre of gravity, and the theorem of rotation of
a rigid body. But in the last analysis, when there have been forces
of constraint which have not annulled one another in pairs, the
setting up of the problem has involved explicitly any unknown
forces of constraint, as well as the known forces, and the former
have then been eliminated analytically, anew in each new problem.
We turn now to methods whereby, in certain important cases,
the forces of constraint can be eliminated once for all, so that
they will not even enter in setting up the equations on whose
solution the problem depends. Moreover, we introduce intrinsic
coordinates and intrinsic functions. The intrinsic coordinates are
a minimum number of independent variables whose values locate
completely the system. They are often called generalized coordi-
nates, and are denoted by q lt qm The intrinsic functions are
,
.
the kinetic energy, the work function or its negative, the potential
energy, and the Lagrangean function L. These we have called
intrinsicbecause they do not depend on any special coordinate
system, or on any special choice of the q's. Later, we shall consider
intermediate cases in which the number of q's, though highly re-
stricted, is not a minimum, and in which, moreover, the unknown
forces, or constraints, have not been wholly eliminated.
* We
choose, in general, the letter m
to denote the number of the q's. But we
replace it by n in these early examples to avoid confusion with the m
that refers
to the mass of the particle.
297
298 MECHANICS
time, t. For example, let a bead of mass m slide freely on a
smooth circular wire,which rotates in a horizontal plane about
one of its points, 0, with constant angular
.
oscillation.
a fixed line and rotate about it. Here n = 2 and q lf q are the
coordinates.
Each two examples may be varied by causing the
of the last
line to move an altogether specified manner. Then, beside
in
ql
= q, or q 1 and q2 the time, t, would enter explicitly.
,
(1) x=f(q,f), */
= *>(<?, 0, * = *(</,*),
where the functions f(q, t), <p(q, t), \f/(q, t) are continuous to-
fa 8y dz
V 'dq' dq
(2)
(3) XX + /*Y + v1 = 0,
8y _
Pz8z\ ~
^+ dt* dq)
y>
y
dx .
dy
y "o~ 7
dz
A TT~ i i
*- ~^r~i
dq dq dq
vanishing because of (3), since dx/dq, dy/dq, dz/dq are the di-
rection components of the tangent to the wire.
The left-hand side of Equation (4) can be transformed as
follows. Wewrite :
T =~
tYI
dx dq
x = . .
From (1)
dx ~ " _
dt dqdt dV
or
/\ dx dx
+ Jt'
. . .
(7)
^ITqO
with similar expressions for y and z. On substituting these
values in (6), T becomes a function of q, q, t. And now the left-
hand side of Equation (4) turns out to be expressible in the form :
W
<W <L
d
L-?L
dt dq dq
For, first, we have :
dT = f.dx..dy..ai\
-z-r m[x
\
+ y-~ + z~).
dq dq dq dq/
.V d dT __
2
/d x dx d y dy 2
d2
' 2
dtdq \dt 2 dq dt dq dt 2 dq,
. / . d dx . . d dy . d <
, 1AN
(10)
_
dT _
= m
^
/
\
. dx _j_
.
dy L . dz
^-.
dq dq dq dq/
Now,
dx _
~ d dx
'dq dt~dq'
For, from (7),
f)'V
UJU (/ IF
s)^ /ft IT
~ JU
q
. . C/ JU
- ~ '
dt^j 8q
dV v dU 8U
~> Y = '' ~'
and thus Q becomes :
= 4. 4.
"*"
.
dx dq dy ~dq dz dq
302 MECHANICS
Lagrange's Equation now takes the form :
dt dq dq dq
where q = ;
and
x = aco sin ut a (0 + co) sin (0 + co)
2/
= aco cos to + a (^ + co) cos (^ + co<)
T = + + +
^- ((^ +
2 2
a?) 2co(0 co) cos co
)
orn /^T
1
r = ma 2 (^ -f co + co cos 0),
= ma 2 co(0 + co) sin 0,
d dT 3T /d*B o .
5ay"-8
a " ma
"(3i
+w8i
,
or
g=-.. dn ,
This last is the equation of Simple Pendulum Motion,
d*Q g ^
-& }**'
Thus the bead oscillates about the moving line OQ as a simple
pendulum of length
Z
-
^~co 2
would oscillate about the vertical.
EXERCISES
1. A bead slides on a smooth circular wire which is rotating
-^ = co
2
sin 6 cos B + 2a s n i 0.
at*
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 303
where <p is the angle which the radius drawn to the bead makes
with the radius perpendicular to the axis.
9. A bead
can slide on a smooth circular wire which is expand-
ing, always remaining in a fixed plane. One point of the wire is
fixed, and the centre describes a right line with constant velocity.
Determine the motion of the bead. 8
Ans. 6 = a - .
+
,
10. The same problem if the centre is at rest and the radius
increases at an arbitrary rate.
(1) x = f(q l9 q2 0,
, V
304 MECHANICS
where the functions on the right are continuous, together with
whatever derivatives we wish to use, and the Jacobians
* *
dq, dt* dq,
l '
dt dq t dq l
and, similarly :
dt dq% dq%
\ d (/ 2 uL uLJ d uL uL uU
dt dqi dq l dq dt dq2 ^2 ^2
and we take q l
= x, q2
= r. Then
dT = . dT = n dU = n
-
-r-r mx, ~^~ 0,J 0;
dx dx dx
or
- o> r =-^sm^.
^2
A
special solution of this equation is found, either by the method
of the variation of constants or, more simply, by inspection, to be :
sni
r "
2co
Hence the general solution is
EXERCISES
1. A is rotating with constant angular
cylinder of revolution
velocity about a vertical axis, exterior to the cylinder, the axis
of the cylinder being always vertical. A particle is projected
along the inner surface, which is smooth. Determine the motion.
Use Lagrange's Equations to determine the motion of a
2.
d*r dd*\ md
O/TI PT 7
%TT
- = ma 2
6,
= ma 2 <f>
2
sin 6 cos 0,
- = mga sin 0,
00 Cv Cv
ma 2
S ma 2
(f>
2
sin 6 cos 6 = mga sin 0,
or
Hence
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 307
(2)
= h,
^sin*0
where the constant h is of dimension 1 in the time, [T~ 1 ].
Combining Equations (1) and (2), we obtain:
d2 6 cos0
= AA + BD sm
.
fA\
(4) r --T-T 6,
at 2 sm 3
6
2
2A cosgdfl 2g dfi
dtdP sin 3 dt
"*"
a
* m
.
dt'
<
--;-?- +*
For a further discussion of the problem, cf. Appell, Mecanique
rationelk, vol. i, 277.
EXERCISES
1. Give an approximate solution for small oscillations near
the point of stable equilibrium, 0, using Cartesian coordinates.
Here,
T = % (Eu* + 2Fui> +
Lagrange's Equations now become, since U = :
(6)
(Fu + Gv) - -
(Ev u
2
+ 2Fv uv +G v v 2) = 0.
L = fVEu'
/ VEu'
2
+ 2Fu'v' + Gv' 2
d\
*0
M = /(A),
(8)
i
(#w'
2
+ 2*W + Gv'*) = h, u' = u, v' = *,
. du
= -J-, .
= dv
U* n rl '
V' -7-
ft n
EXERCISES
Obtain the geodesies on a cylinder of revolution. Observe
1.
2 2
d . h sin 6
"
a(b + acosfl)
3
(10)
where a and & > a are the constants of the anchor ring, and A is a
constant of integration.
4. Show that if, in the preceding question, initially 6 = ir/2,
= or,
= A then
a 2 (6 + a cos0) 2 a2
(ID
dt (b + a cos 0)
2
dx (Py dy d^z
2
dt dq dq dq dt dq dt 2
T = ~(x 2 +y + 2
z 2)
and
* = f(q, 0, y = v(<7, 0, ^ = ^(<7, 0,
where these functions arc any functions subject merely to the
ordinary requirements of continuity, then Equation (1) is true.
We turn now to the general case of n particles, m with the
coordinates (a?, yt,Zi) 9 i = 1, 2, ,
n. Let the position of
the system be determined by m parameters, or generalized coordi-
nates, ?!,,qm and the time,,
t :
*i = /ifoi,
' '
', ?m,
2i
= ti(qi,
-
, gm,
where the functions /,-, (p, ^ t, are continuous together with their
partial derivatives of the first two orders, and where the rank of
the matrix a) is m :
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 311
dx l
dqm
a)
dql
"
-^~' -
i
\
r = 1,
(<7i>
' "
(7m, (7i> > qm, 0> and x^ y^ Zi are given by (3). We
have :
From (3) :
_ _ - etc.
dq r 'dqr
Hence
W
(A\ <L?L = V mi (tfxjdxj d^jjdyi d 2 Zj dz
*^ r V V"^ 2
0?r ^ 2
^r ^ 2
07
d dxi d^dy/i
.
d[
dzj\
2 '
312 MECHANICS
On the other hand,
(5)
I =
\
*
dq r
*
dqr
l
dqr<
Now,
^
dt
For, from (3),
dq r dt'
while
d
dtdq r
Similar relations hold for diji/dq r and dZi/dq r . On substituting
these values in (5), it is seen that the last sum in (4) has precisely
the value dT/dq r and thus the relation
,
I. is established.
v
= Xi
(i)
'
ifei, , <7
A) (<7i, , ?w ,
where the functions /-, ^,-, are continuous together with their
derivatives of the first two orders, the matrix
dq l
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 313
/0 ,
(2)
v ( x* x<
2, *< VdP" ft
dx < v ^< j. 7 ^A
^ 4.
^ ^'
r = 1, ,
ra.
(3) "
It thus appears that
d dT
t =
7
then T becomes :
and Q3 <, Q3 <+i, Qa<+2 are now the components of the force which
acts on m Thus Newton's Equations result at once.
t-.
(4)
(5) $(?!,
'
,?m, % , ?m,
= 0.
(6) AM + -- +A m qm + A = 0,
'
ds dO
(8) s - ad = c,
coin on the rough table, and the bicycle.* But when the Billiard
Ball slips, p. 237, the system is holonomic, for the unknown reac-
tion of the table can be computed explicitly, as the reader can
easily verify, in terms of the velocity of the point of contact,
and thus its components are expressed in terms of the time deriv-
atives of the generalized coordinates.
We
are still leaving in abeyance the question of whether La-
grange's Equations admit a unique solution. Our conditions are
necessary for a solution of the mechanical problem, but not always
sufficient. The study of sufficient conditions will be taken up in
17 and in Appendix D.
* Appell, Micaniqae rationelle, Vol. II, Chaps. XXI, XXII.
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 315
v _
,
Xi ~ till
~ v _
Yi "
, d(J
r/r
'
_ dU
"
r 1, ,
m. When this is true, the Q r on the right of La-
grange's Equations take on the simpler form :
ddT_dT^ = Q
(14) Q =
r Qr + &*, r = 1,
-
, m,
in such a manner that the Q' are essentially simpler than the Q r
where irlf ,
wm are any m numbers which satisfy the equations :
an am i
(17)
be u.
d dT dT ~
no matter what numbers the TT, may be. If, in particular, the
Q r and the jr, are subject to the condition expressed in (14),
(15), and (16), then
d dT
(18)
(20)
(21)
(22)
(23) r = 1,
w/ie
(24)
am \
(26)
' ' '
Ctl/x
dT
(27)
(28) ai,^ + + am qm + a, = 0,
= 1, ,/*,
where the a r,, a, depend on the q r and but not on the ini-
,
ft).
,
Let dxi, dyt, dzi be any 3n numbers, and let be m
carried to the point (xi 8x yi +
fyi, Zi dZi). Then tha
l, + +
system is said to experience a virtual displacement (5xi, tyi, dZifl
"
the word virtual" expressing the fact that the actual system
may not be capable of such a displacement, even approximately.
Thus a particle constrained to move on a curve or a surface would
in general be taken off its constraint, and not lie even in the tan-
gent line or plane.
If forces (Xi, Y^ Z$ act on the system, the quantity
A) ,
*,
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 319
fat t
dx l
09i dqm
a)
^
dq l
dZ n
dqm
is m. Let
(2)
where dq } , 3q m are
,
m
arbitrary quantities. If < 3n, the m
toi, 8y i} dzi are subject to one or more linear homogeneous equa-
tions. Thus only a limited number of them can now be chosen
arbitrarily, the rest being then determined.
Consider the actual displacement (Ax t-, A?/*, Az) which the
system experiences in time At as it describes its natural path.
It is :
(3) A?/t
= <pi (q l + At/!, , qm + A^m + A^) ,
t pi (q lf , gm , ^)
if dfi/dt
= 0, etc. But otherwise in general not.
The virtual work has the value :
(4) ,
(6)
jjj(ir*
+17* + *
r = 1, , m, then (4) takes on the simpler form :
(7) r,-
In either case,
(8) W 5
= Q.dq, + ...+Qm 8qm .
then have :
(10) x = f(q ly q2 0,
, y = <p(qi, q 0, 2 = iKfc, q* t).
work W
s due to the reaction will be 0. But the actual work
done by the reaction in A seconds along the natural path will
in general be an infinitesimal of the same order as At.
It is now easy to see how to compute Q l and Q z m case the
surface is smooth. The virtual work of the reaction of the surface
is nil, and so we need consider only the other forces. The virtual
displacement takes place in the tangent plane to the surface,
and we can compute directly the virtual work corresponding
to the successive virtual displacements given by 8q l ^ 0, 8q2
=
and 5ft
= 0, 8q z ^ ;
cf. further 16 infra.
/ d 3T dT
an
(3)
IS fJL.
(4) a ls 8qi + - -
+ am8 8qm
= 0, 5 = 1, , /*,
* It
may happen that some or all of these equations may be integrated in the
form: F (qi, m t) = 0, where F does not depend on the initial conditions;
-, </ ,
ing about a fixed point with constant velocity. The force which
the plane exerts on the disc at the point of contact will do work
on the disc. But the virtual work of this force, when the virtual
displacement is restricted as above, is nil. Thus we are led to
a suitable set of multipliers 7r r , namely, the 5q r thus restricted.
MgeM K Hence
FIG. 144
(3 ) r = + 1)0.
^(02
Lagrange's Equation becomes :
where
U= Mg a sin 0.
Thus we find as the equation governing the motion :
M (a +2
A; )
2
-^
= - Mga cos
or
fK\ d?0 ag
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 323
(8) 0i
= x, 02
= 27, 03
= 0-
(10)
The first two of these are the equations of motion of the centre
of gravity ; the third, the equation of moments about the centre
of gravity.
What Lagrange's method here has done, is first to eliminate
the internal forces between the particles, just as
did in Chapter we
IV, 1, when we proved the theorem about the motion of the
centre of gravity; and similarly, when we proved the theorem
of moments, Chapter IV, 3 and 9.
324 MECHANICS
Let
Qr = + Qf,
where
Qr~s-MB, Q; = /Z + /*S,
Q* = a(sin + cos 0) S + a(p, sin - cos 6) R.
/x
+ 7r 2 + (M sin cos 0) 7r 3
TJ
= sin (0 + 2X), ?r 2 = cos (0 + 2X), 7r 3 = I/a.
'
f
Sq l =-asin (9 + 2\) 8q,
lo) 1
I
dq2
= a cos (6 + 2X) S# 3
(16)
- MX a sin (0 + 2X) + (My + Mg) a cos (0 + 2X)
+ Mk* = 0. '6
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 325
(17) (/c
2
+ a 2 cos 2X) ~~ + a 2 sin 2X + ag cos (0 + 2X)
= 0.
^~
15. Forces of Constraint. A definition of "forces of con-
straint" from the point of view of physics, which shall be both
accurate and comprehensive, has, so far as the author knows,
never been given. They would be included in such forces as
the X*, *, Z* of 9, which disappear from the Q r I.e., Equa- ;
(1) T = i(4p 2
+ #? + 2
O 2
).
dT = dp dq ,, dr
=
Ap~ + Bq~~ + CV
. .
A
Cr;
Hence
326 MECHANICS
To compute <$>, observe that, no matter what forces may act,
they can be replaced by a force at and a couple. The latter
can be realized by means of three forces :
Cj|- (A -B)pq = N.
EXERCISE
Obtain the six equations of motion of a rigid body by means of
Lagrange's Equations.
(i)
- = Qr + a r8 \ 9 r = 1, m,
dt dq r dq r
2) , ,
(2)
* " "
By the point r is meant the terminal point of the vector r when the initial
point is at 0.
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 327
(3) a,i s qi + -
+ ams qm = aa ,
s = 1, , ju,
T = T2 + T,+ T ,
where
and T19 T
are homogeneous of degree 1 or in the q r or van- ,
The form T2 or ,
(4) 2 </*<**
*,./
/tu, , ju w ,
(7 are functions of #i, -, ^ m t. ,
d dT 8T / A
(5) A ir qi + + A mr qm a rl Xj a r/A X M = Br ,
(6) ai,qi -\
---- + amsqm = C a, s = 1, , /*,
_ ""
an aii
Oi,A OmM
lmti + * '
+ Ammtm + Oml^ + ' ' '
+ Omplfc =
=
=
would admit a solution (, -
, w
77^ , , T/ M )
not the identity.
Moreover, not all the fi, , m in this solution could vanish ;
for then we should have :
+ + = 0.
But the rank of the matrix of these equations, namely the matrix
(2), is /x. Hence all the i) l9 , ^ of the solution vanish a
contradiction.
Next, multiply the r-th equation (8) by r r = 1, m, and , ,
add. The terms in i^, 77^ drop out because of the last p. of
,
0, ,
m ;
(9)
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 329
I/*'** Uft.fi.
wo see that the first m of the Equations (9) admit the integral
given by (3), or :
(11) qa
= Ka , a = + 1,
-
,
m.
The system of m differential equations (9) is now seen to be equiv-
alent, under the restrictions of the dynamical problem, to the
system :
dq8
'
f ,
' ' ' ' *
j\ 1 - - '
(12)
d Ka
, Km, 0, tt = /* + 1,
Here is a system of 2m p differential equations of the first order
for determining the 2m M unknown functions q r K a Their ,
.
s = ad + c.
330 MECHANICS
18. Equilibrium. Let a dynamical system be given, with n
particles m,, the motion being subject to Newton's Law :
(1) Z it 1, n.
(2)
= 0, Y =
t 0, 0,
(3)
= 0,
= 0,
= 0.
form circular disc has a particle attached to its rim. The disc
rests on a smooth ellipsoid and a rough table which contains
two axes of the ellipsoid. Find the positions of equilibrium.
More precisely, the system shall be capable of assuming the
positions defined by the equations :
= fi(q\, ,?*)
(4)
(5)
s ra.
Observe that this last requirement does not imply that m has
the least value for which the x iy y^ Zi can be represented by equa-
tions of the form (4), satisfying the above requirements. It is
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 331
(6) Ft (ft, , qm )
= 0, j = 1, , p < m.
The motion of the system is, first of all, subject to the equa-
tions :
(7) [71, = Qr ,
r = 1, , M < m,
where by definition
\T\
1 lr
=<L2L-VL.
dt dq r dq r
(9)
(10) Qr = 0, r = 1,
- - -
,
m.
If there are relations of the form (8), it may happen that the
Q r can be split up as follows :
(11) Qr = Qr + Q?>
where
(12) Q*ft + + Q*m *qm = 0,
(14) <?>/! + -
+Q'm dqm = 0,
Suppose conversely that (12) and (14) hold when the 8q r are
subject to (13). Then the system is in equilibrium. Suppose
the statement false. From (12) and (14) it follows that (15) holds,
provided (13) is true, and hence from (7) it follows that
(16)
= 0,
i)[r]r5<y r
T~.\
(17) qr = cr ,
r = 1, , m,
follow, namely :
Hence
f)T
= Ai r (ji + ' ' '
+ A m rqmj
v(]r
d BT ....+
= Air .
+ AA mr qm + ..
,. .
, . .
,
- -
terms in -
qm ),
^ -QT q\ (q l9 ,
and so initially
m m
= + A mr cn )c =
^
-!
[T] r dq r
r=l
^ MirCj H r
^ A pc
a,/3
a a cft .
LAGRANGE'S EQUATIONS. VIRTUAL VELOCITIES 333
quadratic form, can vanish only when all the arguments are 0.
We can state the result as a
THEOREM. A necessary and sufficient condition that a dynamical
system, the motion of which is governed by the equations :
_rf d_T __
8T - 1 '"-> m;
V" r
(20) dt'dq r Wr~
i7i + + Om. (7m = 0,
= 1, ,/*< m,
where the rank of the matrix of these last p, equations is /*, be in
} , ,
//, in particular,
(23) Qr = Qr + Qr*, f = 1, , m,
and ?/ # is known that
(25) QlSft + -
+ Q'm dq m = 0.
T =
^ (202
+ ^2 + 2 ^ cos fc,
- 6
)),
.
9X FIG. 146
U =- mga(Q* + ^) + const.
334 MECHANICS
Equations an important class of problems
(1) arc typical for
in small oscillations of a system about a position of stable equi-
librium, the applied forces being derived from a force function,
(7. Let T and U both be independent of t\ let q r = for
r = 1, m, be the position
,
of equilibrium, and let T, U be re-
r, s = 1,
-
, ra, where the coefficients o r, = a ar ,
b r, = b, r are
constants and each of the quadratic forms is definite.
T = q(* + + q'm \
U =- n2 t ft'
2 - - - - nm 2 ^ 2
.
g $-V* r-l,...,.
Their integrals take the form :
(7) q'r
= C r cos (n r t + 7r), r = 1, ,
m.
r = 1,
-
, m,
(10) (bn
- n*arl ) Xi H
-----h (b m- n 2 a rm ) \ m = 0,
r = 1,
- -
,
m.
2
n an
(ii) =o.
6mm ~ n a
2
If the nr 2
are all distinct, they form precisely the m roots of
2
thisequation in n Moreover, each n? leads to a unique deter-
.
following :
f
0j
= cos fat + 7j), f
2
= cos fat + 72) ;
i
^ = A/2 cos fat + 7i), I 02
= ~ ^2 cos fat + 7 2 ),
where
n* = (2
- V2) 2 n2 2 = (2 + \/2) ? -
Cv C*
EXERCISES ON CHAPTER X
1. smooth wedge rests on a table. A block is placed on
A
the wedge, and the system is released from rest. Determine
the motion.
2. Two billiard balls are placed one on top of the other, on
a rough table, and released from displaced from the
rest, slightly
4. A
uniform rod of length 2a and mass 3m can turn freely
about its mid-point. A mass m is attached to one end of the
rod. If the rod is_rotating about a vertical axis with an angular
12. Work the problem of 19, p. 333, when the particles are
not required to move in a vertical plane.
13. Two equal uniform rods arc hinged at one of their ends,
and the other end of one rod is pivoted. Find the motion for
small oscillations in a vertical plane.
14. The same problem when the rods are not restricted to
lying in a plane.
17. A
bead can slide on a circular wire, no external forces
acting. Determine the motion in two and in three dimensions.
Begin with the case of no friction.
* Routh, Elementary Rigid Dynamics, p. 329.
t Appell, Mtcanique rationelle, vol. ii, chap. 24, 446. Many other problems
of the present kind are found in this chapter.
CHAPTER XI
HAMILTON'S CANONICAL EQUATIONS
1. The Problem. The problem of this chapter is the deduc-
tion of Hamilton's Canonical Equations :
dqr_m
" '
dpr__W '
i' ...
' '
dt dp r dt dq r
d dL-- dL = n =
:r; Q~ ^"~ 0, r t
1, ,
m.
cftd</ r dq r
A)
d ML _
dt dK r dq r
r = 1, <
, m, where
(1) L = L(q l9
'
, tfm, *i,
'
, Km,
= i(0r, *r,
(2) L = T+ U,
338
HAMILTON'S CANONICAL EQUATIONS 339
where KT = qr and
T = T(q r , ?r ,
/\ d(L 1? -
,
Lm )
(6)
0(*i,---,O'
where
/ M \
(4) Pr
= 8L r = t
1, , m,
^-,
the f/ r going over individually into themselves. The system of
2m differential equations of the first order in the
dependent 2m
variables q r K r namely, Equations A), thereby goes over into
, ,
fr . dq r 811 dp r 8H - -
(5)
-*=& -*=-*? r 1 '-"-^
where II = H (q r, pr , t) is defined by the equation :
H = ~ L,
(6) I) prKr
r=l
<"
T: r
- r
340 MECHANICS
Let G (y
lt , yn) be defined by the relation :
r=l
where x l9 ,
x n are the functions of y lt , y n defined by
1
. Then
(3) xr r- !,...,.
-%,
For, differentiate (2), regarded as an identity in the independent
variables y lt yn ,
:
dx s
'
d(x u
- -
,
x n) d(y l9 , yn)
or:
, G.) 1
3(*i, , *n)
In particular, then,
/r v d(Gi. ,
Gn ) r=
r$(ri, , /n)"T~
f^i I I
condition:
T
^ .
11
2/r
- dF ,
r = 1
1, ,
w
n.
HAMILTON'S CANONICAL EQUATIONS 341
T~
1 l . xr - r = 1
l, ,
n
n.
^,
Moreover,
' ' '
d(Gi, > Gn) ,
Q
^(2/1, -,2/n)
In particular,
(7) F(xlt ,x n )
r-1
where
3F dG
+ =
(8) J'ft; a?,,..-, x^ Oft; yu , 2/n) Jxy
T=l
r r
=
But
y r ft & ft'
'"' L
w n(if
V \ K li )
K
if
m/}
9
(2) pr = -, r = 1, ,
m.
(3) L +H= %p r Kr .
(5; + ~ t
r ,--, m.
g^
It is also true that
'
dt dt
dq,
~dt
d <9L _ cuu
dt d* r dq r
m
\OJ
dqr
,
- m
~
,
r-\1,, m
/ III.
From the second, combined with (2) and (5), we infer that
(9) -37-
= -r- ,
T = 1, ,
m.
HAMILTON'S CANONICAL EQUATIONS 343
dqr_9H
~ >
dpr__3H
- >
T ~ A' ' '
dt dp r dt dq r
dq r
- = = l,...,m.
.
r
The second equation (10), combined with (2) and (5), leads to
the equation :
d dL _ ~^dL _ 1
~7i ~n t
f 1>
" " "
)
W"
dt 8K r dq r
depending on q r t
alone, and furthermore that the kinetic en-
,
(11) L = T + U.
The transformation :
8L
*-aT r
now becomes :
344 MECHANICS
Thus
r =l
?,,- 2 r
,j-2T.
" c/ r
Hence
(12) ff =
J) p r *r -L = T-U.
T
dH m
as is seen at once by differentiating :
*H = VM^: + T?dH_dpr , BH
dt $, ^ dp
d(Ir dt r dt
"*"
dt
'
(14) H = h.
CHAPTER XII
D'ALEMBERT'S PRINCIPLE
1. The Problem. The general problem of Rational Mechanics,
so far as it relates to a system of particles, can be formulated :
Appendix D.
Two extreme cases may be mentioned at the outset. First,
each variable Xi, Y iy if
Z may be given as an explicit function
of the Xi, yi, Zi and their first derivatives with respect to the
time, and t :
X, = */ fa, y t z if , Xi, y iy z {, t)
345
346 MECHANICS
whatsoever. Then it is seen at once from A) that the following
equation is true :
-i
(2) Xi = XI + X!, Yi = Y( + Yl Zi = Z( + ZI
where the two new forces, namely, the X'it F, Z\ and the
X* t F|*, Z*, are simpler than the old for the following reasons.
i) The Yi, Z{ are either explicit functions of the x^ ?/, Zi,
X't,
known functions arising from forces which are not given as func-
tions of these &n + 1 variables.
+ + =
(3) 2) XfdXi Yfdyi ZfdZi
!--=!
for all values of the 6xi, 5t/, dzt which satisfy the \i equations :
" * "
"11 -"nl
'
(5)
is /x ;
and conversely.
By meansof- Equations (4) the 3n quantities X*, Yf, Z* can
(6)
-
2) (ft
i-l a-l
D'ALEMBERTS PRINCIPLE 347
H M f-
(7) X* 2^ Aia \a , Yi
a= 1
= X( + A ia \ a = +
(8) mix* 5) , nnyi Y( J fi ia X,
W s
-
2) X { dxi + Yi By* + Z, dZi
=i
and of the most general virtual displacement 5Xi, dy^ 6z if whereby the
virtual work of the force Xf, Yf, Z* vanishes, this virtual displace-
ment being expressed by (4); finally the elimination of dXi, dy if dz ir
and X*j Yf, Z*, as above set forth, whereby 3n n equations free
from these unknowns result; this is the spirit and content of
d 'A lembcrt's Principle .
Appendix 1).
i?m,
(D , <im,
i tfm,
dT 8T
ZdtWr'Wr
where
and
(4)
Now, the m
quantities 6q l9 dqm are wholly arbitrary.
,
Hence
the coefficient of each term in (3) must vanish, and so we arrive
D'ALEMBERT'S PRINCIPLE 349
,,, d 8T 8T
Hence
r = 1, ,
TW.
subject to internal forces such that the action and reaction between
any two particles are equal and opposite and in the line through
the particles :
X*
ij
~
__
dXi = + pz - yiji
a t
fyi
= b + yxi aZi
18
2) (x
n n
m g =
)
*
2J
(1)
^= F
dt
/i\
(1)
d dT
5a --" dT n r-i,..-,*. i
ddT _dT ~
_
(2)
(2) [ ^
D'ALEMBERT'S PRINCIPLE 351
then
Qr = Q'r + Q],
where Q'r is for some reason simpler than Q r and where, moreover,
provided
(5) daidqi + ' '
+ a>am8qm = 0, a = 1, , /I,
Qr 2L a <*r^a, r = 1, , W,
d dT dT ,
A _
condition (4) means that the virtual work of the forces which
lead to the Q* is nil, provided that the virtual displacement cor-
responds to the condition expressed by Equations (5).
352 MECHANICS
5. Application: Euler's Dynamical Equations. Consider a
rigid body, one point, 0, of which is fixed, and which is acted on
(1) Qi
= 0, q2
= t, q*
= <p.
(3) q
= \l/
sin 6 sin (p + 6 cos <p
r = ^ cos + ^
By d'Alembert's Principle, 4 :
Compute
M = La + Mf3 + Ny.
Then the virtual work corresponding to the virtual displacement
(5q ly 8q 2) 5g 3 )
= (0, 0, dq z ) is seen to be :
Hence Q = N, and we
3 find :
"
The reader But this is precisely the same solution
will say :
Principle.
6. Examples. Consider the problem of the ladder sliding
down a smooth wall; cf. Fig. 88, p. 147. Let us regard this
problem as the motion of a lamina, moving in its own plane.
AH the generalized coordinates of the lamina we may take the
coordinates of the centre of gravity :
?i
= x, q2 = y, qz = 8.
Then
(1) T = %M(fr + p) + pf F 2
,
(2) 2 r
-Q r 8q r
= 0.
([T\ )
r-l V '
Q^ = S5x, Q 2 6</ 2
= (R
- Mg) by,
Q 3 5<7 3
= a (S sin - R cos 6) d6,
and thus Q r is determined. Let
Qr = Q; + QM
where
Q* = = Q3 =
*
a OS sin 9 - R cos
s, Ql R, 8).
we see that
(5) i;
r-l
(m,
V
- $W
'
== o.
Hence
Qi*fc
= (S
~ R) x, Q 2 6q 2 = (R + S - Mg) 8ff,
Let
Q = r Q; + Q;,
where
Qr = S - /, Q* = B + MS,
Q* = aS(sin + M cos 0) + aR(n sin cos 0).
equation :
f
(1 +M 2
) fyi
= a [ (1
- /i
2
) sin - 2M cos 0] 6q^
I (1 +M 2
) 5^ 2
= a [- 2M sin + (1
- M ) cos 0] 6q
2
3
D'ALEMBERT'S PRINCIPLE 355
- M
2
) a* + (1 + M *) JJ +
2
)
((1
ag [2/x sin 6
- (1
- /*
2
) cos 0].
V+
j
t/) dt,
(T + U) dt = 0.
C : Xi = Xi(u), a ^ u g 6, i = 1, , n,
Xl
, _
~ dxj(u) '
du
Thus a path T in the (2n + l)-dimensional space of the arguments
of F is determined.
By a varied path, F', is meant the following. Let C f
be a curve
in R defined by the equations :
where
a?t(w, 0)
= Xi(u),
'
~~ , e) //
'
\
'
-i , , ,
du
be chosen as the values of the #(,-, #. The curve F' in the
dxi = IH(U), i = 1, , n,
dude >o
Thus
Hence
-r--oa;
.
= A
o-j
<
du du
It is now natural to lay down the further definition :
(4)
358 MECHANICS
8
"=()..,
where Xi and xj on the right-hand side are set equal to x (u,
and ZI'(M, e). Hence
(6)
^-|(J>' + I
It is obvious that
8(F + $) = 8F + d$;
where
, ,*,,, x'n, u), k = 1, ,
m.
/ON *
dF d8F
a
(8)
5?-"5T
And similarly,
6 b
*
FSd* =
(9)
J
a a a
physics.
Variation of an Integral. Consider the integral :
/ F(Xi,
u ' '
>
f
X H9 X19 ,
Xn U) du,
,
HAMILTON'S PRINCIPLE. LEAST ACTION 359
(10) ^
It follows at once as a theorem that
(ID
8 CFdu = 0,
a
(12) dxi |
. = in (a)
= 0; &e< | Ussb
= ^(6) = 0.
For:
d ( $F ^ \
5X " dF 8Xi f d
+ dH W< 8Xi
i *
^ '
d^ V^J V ~d$
Hence
03)
(15) d
jFdu= j8Fdu.
360 MECHANICS
The limits of integration may be varied, too. Let
6 b
where
.
A* = z(o), A'<
= x(a), <
EXERCISE
Since
l/, 6)
f.
du
,eM> d
5 -j- = -7- 6$.
aw aw
ti
>
,-",xn ,()dt,
(1)
jF(x lt ---,x n ,x l
HAMILTON'S PRINCIPLE. LEAST ACTION 361
Xi = Xi(t).
But the symbol written down as the integral (1) is taken in Physics
to mean something totally different. Let
(2) t = t(u), ^u
be any function of u such that, in the closed interval (0, 1),
0<f
du
-f.
(4) dJ = Cd(Ft')du,
o
8J = fbFt'du+ CF8t f
du.
U = T(M),
and
dSt T'(U)
dt t'(u)'
m -
where
8
/z,'\
(7)
= -
t'5x't -
?5
xW .
,
*,
= d
Tu $x<,
..,
st =
^
d
a.
We see,
'
then, that
d dXi
/o\
(8)
*-
5^* = - -
ir
The two formulas, (7) and (8), show that 5i is not invariant
of the independent variable. Why should it be? Similarly,
the variation of an integral is not invariant of the variable of
integration. Much of the confusion in the literature arises from
losing sight of this fact. The " variation of the independent
variable" is supposed to cover this case. It does so when and
only when it becomes identical in substance with the above
analysis.
Hence
n
8T =
(2)
,[*,
5
() + fr S () + 6
a, w.
Variation of the Integral :
(4) J =
Xi = Xi(u), ^ u ^ 1,
and let
Hence
*t <i
tl
< . i <o J
364 MECHANICS
Finally, then :
/i ^
*, /t <i
and
'' /!
n
4. Virtual Work. By the virtual work of the forces Xi, F,-, Z t-,
(1) Xi = Xi(u), yi
= y(w), 2.
= ti(u), UQ-^U ^ u ly
(2) W s
=
(1)
Xi, Y{, Zi are any forces whatever. If these are the total forces
acting on the particles, the right-hand side of (1) will vanish since
each parenthesis vanishes by Newton's Law, and we shall have :
t\
+ 2T ~ + W ) dt =
<
(2) t 0.
J(ST
h
(5) W 8*
= X;txt + Yt8y { + Ztdz< = 0,
t=i
when the variations dX{, 8yi, dzi are chosen subject to certain
conditions. Then Equation (2) takes the form :
(6)
where now dxt, 8yi, dzi satisfy these conditions, dt being still wholly
arbitrary, and
(7) W9 =
2) XJ
t=i
(8)
366 MECHANICS
where A ta Bta Cia are functions of x,-,
, , /<, Zi, x it y f z it
, t, and the
rank of the matrix :
An Cni
(9)
* * *
* 1 I/A v> n/A
(10) Vi
~ '
<?i(qu i q*n,
Zi = ^i(q^ , qmj t)
(11)
*. _ ^ ^ . .
fa *
+ I*-'**.
/1O\
(12) Q = ''O
/^ > r
(
(
V ^i +
A< -r
I V/
Yi
^2/
-r + I
^
/ r = 1, ,
m.
Then
(13) W? = Q! & + +Qm Sqm .
7
and 5T as given by 3, (2), becomes the variation of this latter
ti
where
Let us see just what this means. First of all, the equation is
true under the hypotheses which led to Equation I. These were,
that the path is the natural path of the system, given by the
equations :
where (0)
= t , t(l)
= t lt and the variations
8q r
= ir(u), 8t = / (w),
I
rj r (u) |
< A, | j (u) <
r | h, r = 0, 1, , m,
where h is a definite positive constant.
Furthermore, in Equation X), 8W = W$ is not in general the
variation of any function of q lt qm t. The Q r have definite , ,
values at each point of the natural path, and so are definite func-
tions of u; but they do not in general have any meaning at a
(17)
where
^SfV
^
\du du du du \du
/27
are set equal to the functions (16) which define the natural path
and if dq r 8t are chosen arbitrarily, subject merely to the general
,
Qr = Qr + Qr*, r = 1, , m,
such that
QiiQi + + Qlfym = 0,
provided that
a*i8qi + + Oamfyw = 0, a = 1, , /i,
HAMILTON'S PRINCIPLE. LEAST ACTION 369
where a ar = a ar (q l9 , qm q lf
, , qm 0> an d the rank of the matrix
,
:
(18) 5W = +
but only m /i of the &/ r and dt, can now be chosen arbitrarily.
,
(19)
II.
where the (Xi, F<, Zi) of (19) is the total force acting on m,-,
Equation II., does not assert that the integral of some function,
or physical quantity, is a minimum, or even stationary :
5 I (something) =0 or Id (something) = 0.
t(u, c)
= U, t
Q ^U^ t lt
then
8t s
and the Fundamental Equation I. becomes :
(1)
* In its
leading ideas this treatment was given by Holder, Gottinger Naehrickten,
1896, p. 122. Unfortunately Holder felt impelled to defer to the primitive view of
"
variations as infinitely small quantities" in the sense of little zeros, i.e. infinitely
small constants or functions of Xi, y lt z, t. In the foot-notes on pp. 130, 131 the
"neglect of infinitesimals of higher order" renders obscure in fact, vitiates
the treatment, so far as clean-cut definitions go. The writer cannot but feel that
the inner Holder would have preferred such a treatment as that of the text, but
that he did not have the courage to break with the unsound traditions of the little
zeros, for fear of losing his clientele.
HAMILTON'S PRINCIPLE. LEAST ACTION 371
*i
(3) C(ST = 0.
(4)
to
ti
(5) * + U) dt = 0.
f(T
h
is stationary:
(7) d I (T + U) dt = 0.
d dT _ __
dT ~ dU ______ rp . ,
1 , , . i*jj
' ' '
dtdq r dq r dq r
*>(?!>
' ' '
> ?, = 0, a = 1, , Ml
exist ;
cf Bolza, Variationsrechnung, p. 554.
. The most general
case is that of a system having a Lagrangean Function, or kinetic
potential, L. In the above cases,
L = T + U.
(8)
fldt
to
be stationary ;
or that
i
bCldt = 0.
V
It will be shown in 14 that Hamilton's Integral (8) is actually
a minimum for a path lying within a suitably restricted region ;
but the minimum property does not necessarily hold for un-
restricted paths.
depend on t :
HAMILTON'S PRINCIPLE. LEAST ACTION 373
T = T(qi, , qm q iy , ,?m).
Thus we have :
ti
<0
j(0, c )
= <(0, 0)
= < = const. ; i(l, 0)
= ^.
(3) T = U + ft,
(3') T
[ ?r (u, ),
^|] = U [ ?r (u, c)] + A,
trarily.
374 MECHANICS
Let
(4) SF = -
u
where q r (u, e), t(u,e) are restricted by the relation (3), i.e. (3').
And similarly :
b = iFdu
J\Fdu
(5) -j- ,
0*J e-O
(6.) bT = 8U.
by (3')-
r,
5/T /'\
T
0(1 1 ) = xT 7 'It f
01
i rn jj^'
-f- 7 Ot,
(8) I d(Tt')du = 0.
o
Hence, finally :
(9)
= 0.
lj*Tdt
(10)
subject to the hypothesis that all varied paths fulfil the requirement
that
(11) T = U + h.
In addition^ the variations of the coordinates shall vanish for t = t
and t = Jj.
The Principle thus formulated presents a Lagrangean problem
in the Calculus of Variations with variable end points and one
auxiliary condition :
CTM = o,
(12)
(14) X =- -J-.
* Cf. Bolza, Variationarechnung, p. 586, where the case is considered that there
are, in addition, relations between the coordinates, not involving the time.
376 MECHANICS
Hence the q r (t) are determined from the resulting equations,
8T . dU d dT
(16)
fft'du,
r. s
=
(18)
/
to
/2r '
Let
(19)
or
(20)
Tt' =
HAMILTON'S PRINCIPLE. LEAST ACTION 377
f-
We now have before us a problem in the Calculus of Variations,
of much simpler type the simplest type of all, considered at
the outset. It is the integral (21), formed for the functions
EXERCISE
Show that Equation (1) under the restrictions named can be
thrown into the form :
<i
dT ,
dU d dT\ ,,
=
-x- h o 17 TT-) &Qr dt 0.
O(l r 0q r (It
G(j[r'
(2)
378 MECHANICS
The time is given by the equation:
(3) T = U -
or
r
(4)
vir+T
where
(5) =Vrdt.
We can give a direct proof as follows. The integral (1) has
the value :
(6)
JVu + hVSdu,
where <h(u , q m (u) are arbitrary functions. It is to be
stationary :
(7) d I VU + h VS du = 0.
T = 1, ,
m.
Hence
. S d /vTT+1
/
'
du\
given by (4).
It follows furthermore that
dq r \du/ dq r dq'r du dq r
/i i \' .__ .
\ J4 O^, O^. Q
at oq r oq r oqr
HAMILTON'S PRINCIPLE. LEAST ACTION 379
(2) f\ r*0'*du
380 MECHANICS
must be made a minimum. Set
=
F (r,
>(? + +
'2
^F - ^/M r*0' _
"
W'-Vr + h
. .
(6)
Vr'* + r*0'*
If c = 0, then
= const.
Hence
, . cdr
e
*/;rVhr 2 + r - c2
1
=
""
r
Then
*=+/ v
/T
Al
i
~| /it/* C 22
1L
_ 1 e cos (0 y) = K
EXERCISES
1. Discuss in detail the case c = 0.
* It is true that the interval for u was (0, 1) ; but it might equally well have
been an arbitrary interval : a ^ u ^b .
HAMILTON'S PRINCIPLE. LEAST ACTION 381
(1)
fldt
to
is a minimum for the natural path, provided tQ and ^ are not too far
apart.
The Lagrangean function :
, ?m, ft,
(3) 2
1,1
(4) H+ L = J Pr q r,
T
where
(5) Pr
= r - 1, , m,
J|
and the Hamiltonian function
*
Oarathfodory has given a proof of this theorem Riemann-Weber, Partielle
:
8 I L dt = 0.
fL
The extremals are given by Kuler's equations :
(9)
dqr^M
~ '
dpr___M
* '
r ~
-i1 '
...
'
m '
dt dp r dt dq r
The latter can be solved by means of Jacobi's equation ;
cf. Chap.
XV and Appendix C :
/im
(10)
W + uf
_ .
ff 1
dv
,...,,m ,_ ...,_
dv
(,
Q
as follows. Let (q r , pr ,
t )be a point in the neighborhood of
which Equations (9) are to be solved. A solution of (10) :
(11) V = S(q l9 -
, qm , !,, , t),
(12)
^(ll
*
,
f ^ )
* The existence theorem in question follows at onoo from the theory of character^
istics as applied to Equation (10). That theory tells us thut there exists a solution
of (10) :
V -Stai, -, Q m 0, ,
S (qi, -, q n
-
to)
- , <f> (qi, , qm ).
Here, <p (q\, qm ) is any function which, together with
,
its first derivatives, is
continuous in the neighborhood of the point (qi, qm ). , Such a function is:
<P (Ql,
' '
'. Qm) = Sr7r,
r
where the oti, ,are m
arbitrary constants, or parameters. The function *S
thus resulting is the function required in the text.
If, as we may assume, the function // (q r p r is analytic in the point (q r , , ,
p r Jo), and if, as is here the case, <f> (q r ) is analytic in the point (g r ), then the
,
fundamental existence theorem of the classical Cauchy Problem, formulated for the
simplest case, applies at once, and the theory of characteristics is not needed.
HAMILTON'S PRINCIPLE. LEAST ACTION 383
/-<o\
= SS ~ 8S = l,.-.,mt
(13) pr , fir r
Wf -^,
are satisfied the first set, when (qv, p r J ) are given, by the ,
. f ?r
= /r(i, , m, ft,
'
, ftn, t)
(
I p = r grfai,
' ' '
, OW, ft,
' ' '
, j8m,
05) ^ r
= ^ f,
r = l, .-.,n.
Moreover :
:
Qr
= qr (t), r = 1,
-
,
m.
Let P! :
(q r
l
, t\) be a second point on <~ near by. Connect P
and P! by an arbitrary curve
C: qr
= qr (t), r = 1, , m,
and let q'r (t) = dq r /dt. The curve C shall, however, be a weak
variation,
l?r(0-tfr| <>7,
Let
L = L(q r ,q'r ,t).
values they have for <~ but let ft, ftn have new values, ; ,
not enter explicitly in these equations, and so the fact that they
depend on t does not complicate the equations.
We now apply Equation (18), setting q r = q r giving to q r ,
L = L(q + - + E(q
(21) r, qr , t) % L^r
r (q rj q ry t)(q'r q r) r , q'r, q r 0-
,
(22)
+ S, ($ - L (g =
(23) 2s r
, W" 9r r , (/ 0.
HAMILTON'S PRINCIPLE. LEAST ACTION 385
= - qj + L(q
^jj^ 2) S, r (q r t)(q'r
, r, q r> f).
the first two terms on the right of (21) have the value dS(q r , t)/dt,
and (21) can be written :
(24) L = -
+ E(q r, q'n qr , f).
to
ti
CL <u,
taken along the natural path of the system. For, along <~ Equa-
tion (16) says that
r> Q
~
and so
ti
t.
= 5fa ,0 r
/Lett
to
But in the end points, q r (t) = g r (0- We thus arrive at the final
result :
i i
to to o
(1)
(2) T=
The varied functions, q r (u, ), are arbitrary, subject merely to
the condition that dq r in each end-point, t t t^ It is ,
obvious that the integral (1) lias the same value as the integral :
rt
(3)
JTdt,
(4) T = U + h.
This condition shall hold for the varied paths, too. Thus q r (u, c)
is still arbitrary; but t(u, c) is determined by (4). To prove,
then, that the integral (1) is a minimum for the natural path, it
is sufficient to show that the integral (3) has this property, if
(5) L = T + U.
(6) H = T - U.
(7) L = 2T - h.
(8) r
= 7r(w, 0> ?=<(",),
be a varied path. Consider the varied integral. From (7)
t\ *i
-
(9) /2? dl = fl dt + h(t l * ),
HAMILTON'S PRINCIPLE. LEAST ACTION 387
where
Li'
du
Hence
"'
Since q r = q r for u = w , MJ, the first term on the right has the
value :
StorSZD-StorVo)-
Hence
* *i
= S fer 1 - S far +h - + CE dl
(10) /2T df , ?i) , o) (t, t )
to to
Since // is independent of t :
t\
= W(q r
1
)
- W(q r ).
J*2T<lt
388 MECHANICS
Thus (11) becomes:
C2Tdt = C2Tdt
to t
This proves the theorem. For, if C" is distinct from co'o, then
J5, which is never negative, be positive for some parts of the
will
interval of integration, and hence the integral (3), extended over
C', will exceed in value the same integral extended over the
natural path, as was to be proved.
The case U = const, leads to the geodesies on a manifold for
which the differential of arc is given by the equation :
=
ds 2 5} A rs dq dq t y
.
r, s
.. dq r = 8H dp r = 8H
A) r=l '-' m
,
-
df Wr' dT -W,'
The method consists in finding a large and important class of
transformations of the variables (q r , p ry t) into new variables
(q'r , p'r , t'), such that Equations A) are carried over into a new
system of like form :
3H' 3H'
A/ ,
A) W=W
dq'r dpr
dr
=
~W r - 1 '-'
t
w '
H(<li,
' '
, 7m, Pi, , Pm, and H' (q(,
-
, q'm , p(, , Pm, t')
- = -
(1)
/(2 V'M
1"
H'df) J(2 Prdqr
1'
Hdt),
2 M*H = - Hdt
' '
(2) ?' dt
^ Prdq r
(3) f = t.
'
<?'
= q'rfal,
' ' '
iQmiPi,
' '
, Pm, t)
Pr
= Pr(q\y
' ' '
, Qm, Pi,
'
, Pm, t)
with
' '
^(<7?> <?m, P' } , 9 Pm) _^
' ' ' ' ' '
d(9l9 , (7m, Pi, , Pm)
(5)
- p r dq r ) =d*.
%(p'dq'r
r
^\ ^T,
(6) l
CONTACT TRANSFORMATIONS 391
/*\
(7) *
JT
__ ^ <Nf
and it is not until we have said what these shall be i.e. defined
our function that we can speak of its differential.
A transformation we will henceforth change the notation
from m to n :
, f Qr
= ?'(<?!,
' ' '
, 9n, Pi,
' ' '
, Pn)
(
I
Pr
= Pifei,
' ' '
, tfn, Pi,
' ' '
, Pn)
(9)
= r,
Jpfdtf jprdq
r' r
of as simply connected, and F' is. the curve into which it is trans-
(10) -prdq r)
= d*.
2,(prdq'
r
r
thajt chapter.
*i
(1) n
r ,q' {)dt,
j*L(q
to
and the extremals, which are the path curves, given by Lagrange's
Equations :
/o\
(2)
^ **L ^_n r-V-.,n, 1
(3) qr
= q r (t ; qf, , qJ, qf, , tf n ), r = 1, , n,
(4) <?r
= <?r('; 1, ,?, ?1, ,<7n)>
(5) ?r
= ?r(X), tf r
= g r (X), r = l, , n, X S X g X,,
(6)
- C
"
J
On integrating by parts, observing that
d dq r
di~d\'
we have :
Hence
t\
'(\\
()
- C ~ V (^ J d 3L\ fyr,,
CU +| y dL8q r
J $ \Wr dtWr' 8\ -f 2q r d\
dL
(7)
Thus
(8)
(9)
Hence
0,
X
394 MECHANICS
and so from (8) :
*i
/ k P/y <l
eJX = 0.
(13)
Q
qr = r, Pr = Pr,
=
(14)
/5J Prdq
r
r
/
*J
^0rda
r
r,
, . , , ft,
'
Pr(i, i n , ft, , fti,
3. of the Theorem,
Consequences a) Hamilton's Canonical
Equations. Lagrange's Equations (2), 2 form a system of n
simultaneous total differential equations of the second order.
By means of the transformation (7) these are carried over into
a simultaneous system of 2n total differential equations of the
first order in the (q r p rj 0-space. Let these be written in the
,
form :
(16)
= Qrfop,0, Pr(?,P,0,
*1
d C ^-\ dq r j\ _
-
dX u
(\
dij 2,P'^
'
or
r
=
?1T a\ + pr aTax)
dx '
Integrate by parts :
n
pr
d\
= 0,
*The geometric picture is here slightly different from the earlier one, since
the variables (a ri /3 r) and (q rt p r) are interpreted in different planes. But of course
one may think of a cylinder on T as directrix, with its elements parallel to the
t-axis. On cutting this cylinder with the plane t = t, we have a curve F lying in
the same plane with F. Or, to look at the situation from another angle, t is only
a parameter, and it is the spaces of (a r Pr) and (q r p r) which concern us.
, ,
396 MECHANICS
and we have :
Here,
dq r ~
__ dqr _ n
~ tyr __ dp
~ _ p
" r
Wr>
8t dt dt dt
(Prd r -Q r d r )
= 0.
(19) H = f -Prdqr
'
value of t.
Let the point (a, &, t), for dofinitcness, lie on the extremal through
the point (a, 0', 2 ). Then H becomes a function of (q rj p r 0- ,
If (c/, 0', J
) is replaced by a different point (a", /3", t Q ), the
new H will differ from the old H by an additive term which is a
function of t, but not of (q r , p r ).
More generally, let // be defined by the equation :
(20) II = li+f(i),
dn
CONTACT TRANSFORMATIONS 397
, , dg r _SH dp r _ 8H
(22) r=l,. ..,n.
Hf-Wr' ~dt~~Wr'
First fruits of our theorem. The Hamiltonian Function H
grows naturally out of Equation (14) ; for (18) is but another
form of (14), and (18) at once suggests the definition of // by
(J 9) and (20). Thus if we had never heard of H through the trans-
formations of Chapter XI, we should still be led to it by the
theorem of this paragraph.
f
To
lr- Prdctr) =0,
(<*,$)
(23)
(J5)
(24) f g (Prdqr -
'
Prdctr)
= V (<*, ft t).
('>)
'~
dq r \ __
dV
/o c x
(25)
rj
r
r ,
_
dv
398 MECHANICS
The integral^
on the left is precisely the negative of the integral
(19), or H(q r p r , , f). Hence
(26) H= 2 PrQr ~ ^,
where H is given by (20), and
connected with H
(q r p r t) by the relation (cf Chapter XI,
, ,
.
3) :
(27) L +H = ]T p r q r .
<*> - -
Just as H
was defined only save as to an additive function of t,
so V can be modified by adding any function t and the same y
THEOREM I. Let
r = l,---, n,
qr
= <p r (t ; !,--, or n , ft, , j9)
\
5) Pr dq r
CONTACT TRANSFORMATIONS 399
=
iv) I
^Prdq
'
r I
^LfPrdotr,
r r
.
dq r
-/*_ = dH dp
rr = dH r = ,
"'
i
'
. .
>
T 1
lf
ti
>n '
dt 8p r dt 8q r
9r(a D
' ' ' ' * '
qr , <*n, ft, , ft, t)
a)
' ' ' ' ' '
hr(ot lt , n, ft, , fti,
where r = 1, , n,
Pi,
'
I ^r = A r (a!, ,
an , ft,
'
, ft, O
fee a transformation of the 2n-dimensional (<x r , p r ) -space on the
=
I ^Prdq r
j ^Prdar,
or
(p r dq r
- Prdoir)
= dV (a, /),
2J
r
/3,
C' ^ '
='"~ '
r== >'"> n >
~dt ~dp~r ~dt ~dq r
and conversely.
Equations :
()
,
dqr _ dH
~ '
dpr " dH_
'
1__
r-I,---,n,
~dt 8p~r ~dt ~dq~r
9r
= /rfe,
' ' '
^n, Pi,
'
Pn,
W
/0 , f
1
I
,
Pr
= ,
Vr(q\y
' ' '
,
, ?n, Pi,
' ' '
,
, Pn,
A
(3)
where /f
'
= H' (q' T, p'r t) is some function of the arguments q'T , p'r L
A sufficient condition that (3) be Hamiltonian, i.e. of the form
(4), is that (2) be a contact transformation.
The proof is based on Theorem II., 3 and the fact that the
contact transformations form a group. Let (2), then, be a contact
transformation. Denote it by T. Let (aj, #) be the initial val-
CONTACT TRANSFORMATIONS 401
ues of (?', p'r) for t = tQ. They arise from (a r , r) by T, formed for
* = *o J ^o> let us write it. Thus, symbolically,
(*;, #) = T (a r , r ), or ( r, r)
= T Q-*(cl, ft).
(#, PJ)
= T (g r , p r ).
Finally, consider the solution of (1), whereby the space of the
(a r , p r)
is carried over into the space of the (g r p r ). This ,
D ( r, r)
= (q r , p r ).
On the other hand, the effect of the transformation defined by
the differential equations (3) is to carry the space of the (a'Ty ft)
over into the space of the (q'r p r ). Denote it by A ,
:
(a r , 0J = 57 (;, #)
Next, perform the contact transformation D on the (a r , fi r )-
fer', Pr)
= T (q r , p r)
We have in this way obtained A as the result of three contact
transformations :
A = TDT?.
Hence A is itself a contact transformation, and so the system (3)
isHamiltonian, by Theorem II., 3.
This is the result on which the developments of Chapter XV
depend. It may be stated as follows.
* The inverse of a contact transformation is obviously itself a contact transfor-
mation.
402 MECHANICS
THEOREM. // a system of Hamiltonian Equations (1) be trans-
formed by a contact transformation (2), the result is a Hamiltonian
system (4). The condition is sufficient, but not necessary.
-
(5) ^p r
r dq r 2 ^ dar = dV (a
r
ri r , t),
is characteristic, where
(6)
JT
77 = "V 2 /n
Prtfr
A ~
__
-
dt
(7) 2 M - 2 # da
r
P'
r
'
= dV> ('. '>
is characteristic, where
j jr/
Let
r r
r r
Hence
-
~
dt dt dt'
provided V
(a r $, f) and W(q r p r 0, which arc determined only
, , ,
is a contact transformation if
W
(4\ dqr__<M
dt dp r
'
^Pr__^ff
dt dq r
'
-_!...
lj n ' '
W dQ,_ff'
~ dPr__W
~ _j '
...
'
_
'
<tt P,' rf< 3Q,'
where H' (Q r ,
P r) is connected with H (q r, pr , t) by Equation
(13), 4, or:
(6) H' = H- + (P r Q r - p r q r ).
404 MECHANICS
The (q r , Q r , t) as Independent Variables. Equations (1) repre-
sent 2n relations between the 4n variables (g r p rj Q r r ), and , ,
P
when (q r , p are chosen as the independent variables, (2) and
r)
Pr =
(9)
dW dW dW'dQr
' SW'dq r 8W
dt dt 'dQ r dt ? 8q r dt dt
'
we have :
(10) ff
'
- ff -
OO OO
The Transformation: pr = -^-, P = r ^T' We can write
Oq r Glflr
'
, qn , !,
-
,
a ny t)
&S
/irk\
(12) Pr
= &S , Pr=~ Wr
r>
,
r = l,...,n.
i
Wr
The firstn of these equations can be solved for the Q r in terms
of the (q r , p r) because of (11), and then the P r are given by the
last n equations. Thus a transformation (1) results, the Jacobian
(2) not vanishing.*
The transformation will be a contact transformation, for
(PrdQ r ~ Pr dq r )
=~ dQ r + dq r =~ d3,
(|| J )
(13) H' = H+ -
Vr = fr (Xl, -, Xn), T = 1, , U,
be a transformation having an inverse
Xr = V? r (l/Ii
'
', 2/n), f = 1, , n,
where fr , <f> r are all functions having continuous first derivatives, then
d (y\,
-
-, yn) d (x\, , xn ) _
.
j
-,*) d(yi,--,yn)
d(xi, -
P = r a r then
,
O&
/-.
(14)
^\
p, = v*S
w ,
s^
Qr= W ,
r = -
l,..-,n,
Choose W(q p r, r,
= W"(q r , I\, t), therefore, -so that
S=- W"
Compute dW"/dt and show by the aid of (14) that (6) yields :
(15) ff' = // + .
06) -,
where q r , Q and
r, W are expressed as functions of (TT*, 0-
/)^f OC|
3. The Transformation : q r = -5 ,
P = r -^r- If (p r , Qr) can
Gpr #V r
be taken as the independent variables, and if we set
S = W+ Pr?r,
CONTACT TRANSFORMATIONS 407
(17) ^= d P r'
_
'
(18) H' = H-
Conversely, if S(q lt q n a lt
, ,
a n ) be chosen as before, and
,
Qr = Wr T = n,
X^ Xli
(2)
Wn
is m. Thus m of the Q*'s can be expressed as functions of the
f\ f\ n ff /\ __ 1 /yyj
* ' '
> Qn, <7i, , ^n, can serve as the 2n + 1 independent
408 MECHANICS
variables. But the function W(q r p r , , t), when expressed in
terms of the new variables, does not depend on ir ly ,
ir m :
- p r cfy r) = dW* (q Q
(5) 2) ( Pr dQ r r, r, 0.
I (^-CK -?(*
It is not, however, in general true that the coefficients of the
differentials vanish.
dQu dQm
,
can be eliminated, the resulting equation being
of the form :
Xm+l = 0,
- .
,
Xn = 0, Y, - 0, ,
Yn = 0.
= n
n o
(8)
Multiply the fc-th of these equations by X& and add to (6). Then
determine the X/t's so that the coefficients of dQ dQ m vanish. } , ,
_
~
'
Wr + Wr '
(9)
r
= l
... m
CONTACT TRANSFORMATIONS 409
^ dQr ^ d&r ^ ^r
dtt r
""
dttr
^.
^r ,
^'
dt 2? 3Q r dt ^ dq r dt dt
-- -J
If, in particular, the 12's do not contain t explicitly, this equa-
tion reduces to
(11) H' = H
CHAPTER XV
SOLUTION OF HAMILTON'S EQUATIONS
1. The Problem and Its Treatment. We have considered a
great variety of problems in mechanics, the solution of which
depends, or can be made to depend, on Hamilton's Canonical
Equations
m
(i)
:
dqr - dH dpr --
~ m r-1 '
..-
'
n '
~dt~~Wr' ~dt Wr
where H
is a function of (q r p r ). The object of this chapter is
, ,
Qr = F r (q 19 , q n Pi,
, , P,
Pr = G r (q ly
' - -
, q n , Pi,
' '
, Pn,
Equations (1) are carried over into equations of the same type :
~ '
dt d/V dt
in general equal to H .
dV
--
(A\
(4) +
_i_
(5) V = V(q lf , g, i, , n, 0,
The function (5) once found, the further work consists merely
in and the solution of equations defining the
differentiation
grated at sight :
where a r ,
are arbitrary constants.
($ r On substituting these val-
ues in (2), the inverse transformation,
f qr
= /r(Q!, -,, PI, ,P,0
qr
= /r(i,
' ' '
, n, ft,
' '
i fti,
(8)
Pr = 0r(i,
' ' '
, <*, ft,
' ' '
, ft,
^
,
.
Pr
_ dV pr _ dV _ t
r-l,..-,^
~Wr ~~Wr'
where F is written for the arguments q r , Qr :
= dV dV
pr
Wr'
&T
=~^ r=l,..-,n,
Q = F
r r (q lf , g w p lf
, , p n)
(10)
' ' ' '
, ^n, Pi, ,
412 MECHANICS
such that the new H will depend only on the P rj but not on
Q rj
t. In particular,
#' = P^
= 0, r=l,...,n.
Thus*
Qi
= t + /3 } , Q, =ft, 5 = 2,
P =
r ar ,
r = 1, ,
n.
(12)
(13) 1
ft, ft,
HA\ dW n
(14) pfSS 9 Qr =
Wr
where W is a solution of the equation :
m dW
or:
^'-'^
iff
Tf = W(q l9 ,?n, A, 2, ,)
Here TF depends on the arbitrary constant A, and also, in a suitable
manner, on n 1 further constants, or parameters, a 2
" * '
<** t
Pi = h ; P, = a a, s = 2, ,
n.
(15)
,w
(16) (PrdQr
- Prdq r ) = dW (q r, pr
2}
r
, t),
This the theory the student need know from Chapter XIV,
is all
(2) /r-ff +
f.
The transformed function H' can be made to vanish identically
if we can find a solution V of the partial differential equation :
^.......vj^
3(ll >)
On setting S equal to this function F, and making the trans-
formation (1), H' as now determined vanishes identically. Thus
the transformation :
(5) p,
= Pr r=l,...,n,
g, jfc
where a r is replaced by Q r in F, transforms the Hamiltonian
System (1) to the Equilibrium Problem:
'- 1 '-'"'
t-o- T'--
The solution of these equations is the system of equations (6),
1. These are the values of Q r P
r to be substituted in the
,
transformation (1) ;
i.e. in the present case, in (5) :
(7) p, = r-l,...
0,=~g,
.
g, f
The last n of these equations can be solved for the q r 's because
of (4), and then the first n equations give the pr .
SOLUTION OF HAMILTON'S EQUATIONS 415
Pr
= V r ,
-
, gn , !,, n, Q, r = 1,
-
, n,
EXERCISE
Pass to the Equilibrium Problem by means of the transforma-
tion studied in Chapter XIV, 5, Exercise 1 :
dS n dS
Here,
*'~W Qr =
W r
>
" = V--,".
1
Let V = V(QI, , qn , a lf ,
an ,
be the same function as
that of tho text a solution of Equation (3). If, then, we
replace a r by P r and set S = V, the transformed H '
will vanish :
dQ r = dP = r n = ,
r l,...,n.
~W >
~di-
'
Q =-
r Pr ,
P = r Qf r , T = 1, , tt,
(1) T=
^q\ V -- \q*,
0<X.
Thus
(2) L = T+V-?-f
(3)
416 MECHANICS
al
a;
dq - p dp -- \n
A9
Tt~
'
di ~m'
We propose to solve them by the method of 2. The equation
for determining V, 2, (3), here becomes :
(6) V
(8) F = fl + W,
where 12 = 12 (f) is a function of t alone, and W = W (q) is a func-
tion of q alone. If this be possible, we shall have :
'
+ V= o.
X d!2
obvious that a ^ :
* Let the student disembarasa himself of any fears due to his ignorance of tho
theory of partial differential equations. No such theory is needed in the kind of
application in Physics which we are about to consider it would not even be help- ;
ful in practice. The single function V(q, a) is obtained by a simple device fully
explained in the text.
There is, of course, a most intimate relation between the theory of Hamilton's
Equations arid the theory of this partial differential equation, as is indicated, for
"
example, by the theory of characteristics" cf. Appendix C. The point is, that
;
this theory is not employed in such applications as those illustrated here. For the
latter purpose, a single solution V(q\, q nt ai, , is all that is required,, ,
and such a solution is obtained by ingenious devices of a homely kind, as set forth
in this Chapter.
SOLUTION OF HAMILTON'S EQUATIONS 417
2
J_/dTF\ X
+ 2==a
Q = ai f
(dW\
-j )
W= I V2ma - m\q*dq.
Thus
/
(10) V=-o* +
/
dq
(11)
__^_
fl_-^--<- m - ^_
u
Xg2
and thus
(12) q
p = V2m^ cos -
(13) \- ( 0).
not paid heed to Condition (7) on the one hand or, on the other,
418 MECHANICS
considered that the second equation (11) is restricted. Here
then is a difficulty.* Either we must follow the theory as hitherto
developed, using single-valued functions W
and V ; then t is con-
fined between certain fixed values. Or else we must introduce
multiple-valued functions F, and then we must go back and
revise and supplement the general theory.
There is, however, a third choice a way out, whereby we
can remain within the restrictions of the present theory. Accord-
ing to that theory the solution given by (12), (13) is valid so long
as
EXERCISES
1. Obtain the solution of Equations a) in the form :
(14)
p = V2ma sin ^~ (t
W= I V2m<x 2
m\q dq + C(a),
o
nit
-fi, ^ A.
Obtain the equation for motion near the point of stable equilib-
rium:
dq
f
where t is restricted. Hence discuss the two cases a) oscil- :
p = mq,
p*-
420 MECHANICS
dW = V2ma 2m
+ 2
gq.
Since
_ dV dW
P
dq dq'
we have no option as to which radical shall be taken. If the body
isprojected upward, q will bo negative for a while, and so we must
choose the negative radical for this stage of the motion. At the
2
turning point, (7) V/dqda does not exist.
is not fulfilled, since d
We
have now a new problem, as the body descends. The
choice of W
must be made on the basis of the positive radical.
Nevertheless, both stages of the motion are covered by the solu-
tion for the first stage :
ft 1
2^y
/> ft o\ 2 "%/_____ // ___ /9i
p = mg (t /3) V2om.
Why?
4. H, Independent Reduction to the Form, H' = Pi.
of /.
where
*'*' )
'"'*S * o,
dS ~ dS
fn\
(2) pr =
^, Qr- W ,
r = .,
l,...,n,
H = H(q lf
- -
, qn , p lt - -
, p n ).
SOLUTION OF HAMILTON'S EQUATIONS 421
Then
H' = H'(Q l ,".,Q*,P ,---,P n ). l
(5) H'(P19 -
, Pn) =P x
.
v
,R
(6)
W = W(q ly
-
, g n A, , 2, ,)
depending on n 1 arbitrary constants 2, ,
an and of
course on h, which is also arbitrary such that*
(7)
, 2, , n
(9) S(q l ,
- -
, qn , a,, , a) = TT^, , tfn, A, 2, , )
(10) Pj = ai
= A ; P. = a., 5 = 2, , n,
#' = PI,
as was desired.
dQ.
1' 0, 2,
-dt^
dt dt
(12)
r = o, r = 1, ,n.
Ci
= < + ft, Q> = P; s = 2, -,;
(13)
P =
r ar ,
r = 1, ,
n.
(14) pr = {T ' n,
dP r
we liave :
(15)
+W^ +
(16)
+ +
SOLUTION OF HAMILTON'S EQUATIONS 423
=
(17)
=
h = H(q i9
- -
, qn , p l9 , p).
*That H is here constant along a natural path follows from Chap. XI, 3 :
dH - dH -
'
Tt IT
424 MECHANICS
~^
d(W q ,
- -
,
W Q )
(8') -^ * 0,
3(alf <*2 , , a)
where W= W(qly q* h, a lt , 9 , n ) is determined as before
from (6), and h = $(!, 2 ,
-
, ). Thus the transforma-
tion (14) is justified and Equations (12) become :
dQ r
dt
(120 1, ,
n.
dt
Qr = Ur t + Pr, r = 1, , H,
where
o> r = $r(<*i, ,
a n ), r = 1,
-
, n,
(19) , .
ft,
T= & + 2
ft
2
+ </s
2
), I/
SOLUTION OF HAMILTON'S EQUATIONS 425
= + p + P^
tt
^ Cpl2
*
'
0(a, a,) f
by setting
W. = W, +W + 2 W 3,
where W r
= W r (q r) is a function of q r only. Thus
n
0.
"j i j
Vd^/ Vrf?2 / Vrf
^J = 2m(/l-a 2
2
-a 3
2
)-
JTTf
* -v/O TTZ A /O O Q
- a,*
- a,*)
W 1
= - CV2m(h - 2
2 - a 3 2) + 2m
Cl
Q\
W =- A/2m(A - <*2
2 ~ 3
c,
_
We are now in a position to write down the solution of the
problem. It is given by Equations (15), 4 :
t + ff
=^=- m Jf ^i
-
V2m(h -
'
dh 2
2 2
3 ) + 2m*gq l
A = |^ = 2ma
Jf
a
8.
and hence :
- 2
2 - 2
3 )
ft)
SOLUTION OF HAMILTON'S EQUATIONS 427
s = 2, 3.
descending, p becomes
v positive, and dWl /dq1 = dW/dqt cannot
be expressed by the negative radical. This second stage of the
motion, in which the body is falling, could be dealt with by apply-
ing the method afresh with suitable modifications in particular,
by taking the positive radical for dW t /dq^ But this step can
be eliminated if we observe that the equations we are integrating,
Hamilton's Equations, here become :
r =
di ~m Pr > 1,2,3;
2,3.
dt di
generally.
EXERCISES
1. Central Force, two dimensions, attracting according to the
law of nature. Let q l = r, q2 = <p. Then :
= R
2m\dr
428 MECHANICS
/dR\* n , 2wX a2
=2mA +
.
(W) -^
Thus
W= J
2mh + T -^dr
T
+ <*>,
r*
where either the plus sign or the minus sign holds throughout
the first stage. Hence
r
dr
t + =+
~ m f
J 2mX
Vo2mh
z, ,
H r
02
=
2mh
r r
Discuss the case that the radicand vanishes for two distinct
positive values of r, expressing r as a periodic function of ^>, and
evaluate the integral that expresses t cf. 9. ;
~2m
TT = fl + + *;
2m\
J VO
Complete the solution and discuss the cases that the radicands
have distinct roots.
SOLUTION OF HAMILTON'S EQUATIONS 429
and integrating :
= ef " ^
dh
I
W ff(q,h)dq.
Then
y
_8W
dP
Thus
"- 8q-
6. Comparison of the Two Methods. We have studied two
methods of solving Hamilton's Equations, a) Reduction to the
Equilibrium Problem b), when // does not depend on t, Reduc-
;
(1) V = - ht + W,
where W= W(q lt q n h, a a , , >
*
> n) is the function of 4,
and h, as have been replaced by P P
lt 8. The transformation of
that Exercise,
r)V ?)V
(2) pr = ~ Qr r = 1 n
yields an H '
that vanishes identically. The transformed Hamil-
tonian Equations thus take the form :
W
(Ti
^ r
n
^ r
~'n r
r ~ 1 . . .
n
-dT-' lj *
'
dt
430 MECHANICS
... I Qr = Pr,
= !,--,;
r
(4) 1
I
P, = h, P. = a., s = 2, ,n.
= 8V
(5)
* = 2,
. .
,
n.
But
^ *
3V_ = 3W
~ '
W =_ +
3W
'
3V ~
=
dq r dq r dh dh da, da,'
Here it is :
(2)
f-f-0. -.-.,
and hence
(3) ?>
= ,
5 = 2, ,
n.
(4) h = H(q p l9 19 ,p n );
SOLUTION OF HAMILTON'S EQUATIONS 431
/*\ dW
(5)
where
(6) W = W( qi1 -
, ?, h, a -
,
an )
is a solution of the equation :
l, ,...,
m \
J,
,,,...,
-
and P =
l A, P, = a a, s = 2, ,
n.
To find such a solution we turn to the Method of Separation
of Variables, which has rendered such good service in the past.
Let
'(9) . w=W l + -- + Wn ,
where W r
= W r (<?r) is a function of q r alone and of the n
parameters, A, a2 , ,
. From (5) and (3) we see that
a" 2,
- - -
, n,
and so we try :
TT, = g., s = 2,
- -
,
n.
(11) fl (g 1 2 ,---,n)=A.
,^,a
If we assume that
fiff
(12) l^ffp.feuPi."*'--.*..)* '
^Pl
(13)
432 MECHANICS
for p, :
(15)
= *(?i, A, a,, -,).
-^
Now choose as :
(17) W = V + otf, + +a n qn
2
d' v
" '
r '
M!!
dp, dh
= 0, r-l,-..,n,
ft
= + fc, Q. = A, s = 2, ..-,n;
P = x A, P, = a a, s = 2,
- - -
,
n.
<7i
aw
l9
dh dh J dh
(18)
Pi
= -8W = *Wi,
r , ,
h, a2 , ,
a n),
.
dW =
p9 = -7T-- a,, s = 2, ,
n.
(W =
~^~dqly s 2, '--,n.
/' 8
EXERCISE
Obtain the final result (18) directly from Hamilton's Equations.
(2) p,
= a., s = v + I, ,
n.
(3) W = v + a* +l qr+1 + + an q n,
(5)
a^T
434 MECHANICS
Equation (7), 7, for W now becomes :
(b)
/c \
nL = rj -
, q,, TT-, , a^+i, ,
//^ft, Tjjp
(7)
/Si
= -
+ qi, I = v + 1, ,
n.
g2 = (p. Then
"
2\dt 2 dt 2
Hence
Let
W=v+aq 2 2.
SOLUTION OF HAMILTON'S EQUATIONS 435
1 f/^\ 2
, 2
2
1 ^ _
2m \\dq l
/ (7j
2
J ql
'
or
rt 7 . 2mX <* 2
2
,
2mA H 1-
2
dr.'
r r
where a definite one of the two signs holds for the first stage of
the motion.
Equations (18) of 7 now give the solution of Hamilton's
Equations in the form :
. ^ dv /* Sv .
or
dr
,=/- c 2mh +
2m\
ft
=
Then
Let 0, Then
Pi Ps
436 MECHANICS
2
P
H =
2m
( vI
\
*
+ Si 2 2
cos 2 ?2 /
tf, ?, ql
W = v + ci^s,
where v is given by the equation :
2
_ x
J_[7-^Y + A. fi?Y 4.
*
i = ,
2m lA^/ qf\dqj q,
2
cos 2 q.2 J ql
0.
cos
On setting
v = R +
the variables can be separated :
- 2m\r =
r*(~^ + 2mhr* +
Hence
ft
= rv^ _ a .,
2
where the signs are determined for a particular stage of the motion,
and c, T are arbitrary numerical constants. Adding the further
term a 3 g 3 we have:
,
W = v + ag^, v = R + 0.
We are thus led to the solution of the problem in the form given
by (7), 8:
SOLUTION OF HAMILTON'S EQUATIONS 437
*
,
+A -.A
r
dr
&=- J T, 2raX 2
c
.
9^/0
r2
a,
(2) \2rnh
* -\
r
\
r2
6
d6
- a32 sec 2
v a, 2
2
a32 sec 2
2raX
= 2mA
h =- mav ,
2a'
and hence
= 0.
438 MECHANICS
Thus the condition
T =
Let
tfl
^ Q> <?2
= <P>
Then, since
= dT
we have :
Pi
= Ad,
Furthermore,*
U= Mgb cos 0.
Thus
* It is
necessary to change from the earlier notation h for the distance from
the peg to the centre of gravity, since h plays so important a rdle in the present
theory. Let the distance be denoted by 6.
To
where
The
form:
determine
Irl
2 Li 55?
sin 2 ft
t/
c
^2
-
dv*
^2 =
= r+ V(2^ A La2
_
SOLUTION OF HAMILTON'S EQUATIONS
2 -
v
+ C^ + l(
,
(24 A
we have
1 9 .
L = 4,
v/
:
l/a2 cosq
- La - N cos ft)
2
2
and the sign of the radical must be chosen with respect to the
special stage of the motion under consideration.
brevity,
Moreover, for
+*-*
N
1
- a 3\ 2 1
= 2AMgb.
) J
sin 2 q l
+M
-
,_
.
* 6 COS * -
7,
(
now
cos
,
2 cos q1
?,
-
-
takes on the
L
*
439
3)
2
?1>
,
Thus
4- C-
A sin ft (
'~e/~H v (2Ah L 2
2 ~ A^ cos gj sin 2 gj ( 2 cos gj a s) 2
Let
u = cos #,.
9l
= r ^ cose,
e/
c
+V/<W
where
Chap. VI, 18. But compare the technique. With only Euler's
440 MECHANICS
Dynamical and Geometrical Equations to work with,* elimina-
tions had to be made by ingenious devices, whereas the present
advanced methods free the treatment from all artifice. The
fundamental equation in desired form is evolved naturally, directly,
from the general theory, not untangled from a snarl of equations.
Instead of having to solve three equations for 6, by more or <j>, \j/
7
less ingenious methods of elimination, the functions 7 [7, and ,
EXERCISE
Study the motion of a top with hemispherical peg, spinning
and sliding on a smooth table. Show that
where
-
F(u) =
(2h ^-
11. Variation of Constants.
Perturbations. In the problem
of perturbations the motion which the system would execute if
only the major forces acted is regarded as fundamental, and
then the variation from this motion due to the disturbing forces,
thought of as slight, is studied.
This analysis of the physical problem is mirrored mathematically
by writing down Hamilton's Equations for the actual motion :
___ -.-
''
~dt ~Wr dt" dq r
' '
(2) H = H, +H {
.
* It
is true that in the earlier treatment we had two integrals of the differen-
equations of motion to work with at the outset, namely the equation of energy,
tial ;
T = U h, and the equation arising from the fact that the vector moment of mo-
+
mentum <r is always horizontal. But even so there were three equations in 0, $, <f>
to integrate.
SOLUTION OF HAMILTON'S EQUATIONS 441
fA\
(4) Pr
r
where
Let
V = Vfe, -,?, a lf ", an ,
O / v 7~ VJ.
0(a l9 -
,
an )
f Qr = Fr(p lt
' * '
9 Pn 9 qi,
' * *
9 qn,
,.
I P r = G (p r lt
'
jpniQi,
' " ,qn,
or
f 9r
= fr(P , P, Q -^Qn,
1 pr = g r (P ly
'
,
Pn Q On,
, 19
-
,
(7)
f= 0,
^= 0, r=l,..-,n.
f Qr
= /r(i,
' ' '
, n, ft, , 0n,
1 Pr
= 0r(a lf
" -,, ft,
'
',0,
;
Thus the transformations (4 ) or (4"), and (9), identical
that Equations (1) will go over into new equations of the same
form:
(W)
dQ,_8ir
~ dPr__9ff'
~ j.
dt dP r '
dt Wr ' '
Here H' = H'(Q P r, r , <) has the value, cf. Chap. XIV, 5, Ex. 1,
(15):
(11) H' = H+
But from (6) :
Hence
H' = H-
SOLUTION OF HAMILTON'S EQUATIONS 443
(12) H f
= ff t .
dQr^dH, dP r _ m,
(14)
' ' '
Pr =flTr(P,, ,Pn,Qi, , Qn,
tions (9), are constants the equations there representing the solu-
tion of the major problem, (3). On the other hand, they can be
identified with the variables P r, Qr of (14), these variables being
determined by (13), and then Equations (14) represent the solu-
tion of the actual problem, (1).
We can attain complete confusion of ideas, as is done in the
literature, by changing the notation in (13) and (14) from
Q Pr
r, to r, oL r . Thus (14) goes over into the form of (9), and
(13) is replaced by the equations :
da r _
~ 3(- g,) df)r _
~ B(- H,) -
' '
r L >'"' n >
dt d0 r dt da r
444 MECHANICS
where 1
H H
= 1 (Q r P rj t) is now written as l (ft1) a1) t) the
,
H )
tives are not all 0. Let the Hamiltonian Equations for the undis-
turbed motion, namely, (3), be transformed by a new contact
transformation :
/i/r\
(16)
/
*(,, -..,.)
\
= T
Let
, qn , !,--, a n , t)
*
be a solution such that
- ^ ^
/1>7\
(17)
where
Pr
Wr
, Pr=~r>
W ,
,
/ OS dS \ dS
h qi ,..., qn + -,,
=H( ,~,...,--,t)
where h is an arbitrary constant, and seek a solution :
S = S (<?,, , qn , h, 2, , On, Of
such that
' '
t gt , ',
Qn
d (h, as, -, )
h = <p (ai, , an )
in iS. If d<p/dai ^ 0, this will be the function desired.
SOLUTION OF HAMILTON'S EQUATIONS 445
' '
dt dP r dt 8Q r
' ' '
dQr
" = 0/7, dP _ _ r 3/7, _ 3jp - "' n
r 1 -
obtained from these equations are the solution of the actual prob-
lem, or Equations (1).
*
Carathgodory treats Equations (20) as follows. He writes
X//! instead of 1
H :
f Qr= Ct r + + X C2r +
\Cl r 2
,
< 22 >
( P =
r r -va t+ Wlr + ' '
,
A +B=B+A
AB = BA
A(BC) = (AB)C
A (B + C) = AB + AC
(B + C)A = BA + CA
A brief, systematic treatment such as is here required is given
in the Author's Advanced Cakulus, Chap. XIII. For a first
approach to the subject the Hamiltonian notation of S and V for
the scalar and vector products has the great advantage of clear-
ness in emphasizing the functional idea the concept transfor- :
A = B.
denoted by |
A |,
or by A.
Addition. By the sum of two vectors, A and B, is meant their
symbol 0.
A+B = 0, or B =- A. B
FIG. 149
Moreover, we understand by wA, where m is
any real number, a vector parallel to A and as long its m times ;
aA +r- 6B means
a
A
.
H
,
6
rr B.
_
+ jo
.
a a + ,
o a +o
Vector addition obeys the commutative and the associative law
of ordinary algebra :
A+B =B+A
A+ (B + C) = (A + B) + C
Subtraction. By A B is meant that vector, X, which added
to B will give A :
B+ X = A, X =A- B.
VECTOR ANALYSIS 449
To obtain X
geometrically, construct A and B with the same
initialpoint ; A B is the vector whose initial point is the
then
terminal point of B, and whose terminal point is the terminal
point of A ; Fig. 149.
Cartesian Representation of a Vector. Let a system of Cartesian
axes be chosen, and let i, j, k be three unit vectors lying along
these axes. Let A be an arbitrary vector, whose components
A A A
_
along the axes are lt 2J z
. Then evidently
A = AJ + A 2 j + A z k.
then
A+B = (A, + BJi + (A, + B,)j + (A, + 3 )k.
Also :
A ^
Resultant. If n forces, F x ,
F2 , ,
Fn ,
act at a point, their
resultant, F, is equal to their vector sum :
F = F t +F + 2 + Fn .
If n couples, uM M 2 , ,
M
n act on a body, the resultant ,
M = MJ + M + 2
-
+ Mn .
X = ZA + mB + nC,
where Z, m, n are uniquely determined.
Differentiation. Velocity. Acceleration. Osculating Plane. A
variable vector can be expressed in the form :
A =
450 APPENDIX A
where i, j, k are three fixed vectors mutually perpendicular. If
/(Oi <p(t), ^(0 have derivatives, the vector A will have a de-
rivative defined as
lim-rr =
Its value is :
Moreover,
d(mA)- = dm.A of A
+ m ~T'
.
-~^j -y~
ax ax ax
If a point P move in any manner in space, its coordinates being
r = xi + yj + zk,
the vector velocity of P is represented by
AB = BA
A(B + C) = AB + AC.
ii,
:2 i ;2
j
i
i,
k2 _
K i
i,
jk = 0, ki = 0, ij
= 0.
= AB
Differentiation :
a2 = 1, and aa' = 0.
AXB
"
and is read : A cross B." FIG. 150
452 APPENDIX A
If one of these vectors is 0, or if the vectors are collinear, neither
being 0, the vector product is defined as 0, and these are the
AXB=-BXA.
The associative law does not hold ; e.g. (i X j) X j 5^ i X (j X j).
C)=AXB
and
In any case
A XA = 0,
and so, in particular,
i X =
i 0, j X j
= 0, k X k = 0.
i J
-^-
**l
Differentiation:
- -
dx dx dx
A (B X C).
VECTOR ANALYSIS 453
A -
(B X C) = 0.
v = () X r,
v =
x y
and
vx = zwy ywz
Vy
= Xtl)g Z<l) X
vx = (z
- c) co y
- (y
- 6) wz
vz (y 6) o) x (x a) uy
let (f, ??, f ) be the coordinates of any point P fixed in the body,
the origin being at 0' but otherwise the (, rj, f )-axes may
;
P 7
COf CO, C0
* i r
454 APPENDIX A
Localized Vectors. It is sometimes convenient to prescribe the
initialpoint of a vector, or the line in which the vector shall lie,
as in the case of a force acting on a particle, or a force acting on
a rigid body. It is with reference to such vectors that the follow-
ing definitions are framed.
By the moment of a vector F with respect to a point is meant
the vector
M = r X F,
TI X F! + r, X F2 ,
(r X F),
y
z n^ n% 7i 3 fc HI w2
VECTOR ANALYSIS 455
IS + mS + 1
mS + m + m = 2
2
3
2
1
n* + n2 2
+ n3 2 = 1 n
+ tn w + n n =
2 3 2 a
==
+ m^m + n n =l 3 x n^j + n2 Z2 + n3 i!
3
=
-f w,m + n n =
2 x 2 JiW, + Z2 m +
2 J
3 m s
=
m 3 n2 m = l
n2 Z3 n 3 Z2
n2 = Z3 m I Z^j
Z3 =
1.
APPENDIX B
^
= /()
(dij\
~jj)
where
I. f(u) = (t*-a)(6 -*)
or
II.' /(w)
= (w
- a) (6
- tO
a ^u ^ bj
du
= du
dt
2> a ^u ^ b.
<
'-Ajsr
Geometrically, the function on the right of (2) can be inter-
preted as the area under the curve,
(3)
\ /
w
/r\ .. _ m
V(u-a)(b -u)t(u)
The area under the curve (5), shaded in
the figure, represents the integral (2), or: v
~~
r du_
(6) t
J V^-~Mb~^
Thus this area expresses t and brings out
the fact that t increases as u increases.
a u b
Conversely, u increases as t increases.
FIG. 154
Let A be defined by the equation :
r du^
(7)
J V(1T^~ci)(6~-
(8) u =
FIG. 155
is as shown in Fig. 155. Its slope is at
each extremity and positive in between.
The definite integral, (2) or (6), has now served its purpose.
It has yielded for a restricted interval,
^ t g A,
a particular solution of (1).
Thus
= *(- 0,
-A
458 APPENDIX B
The extended function :
fdu\ =
(w)
Hence the function <p(t) thus defined in the interval ( A, A), or
u = <*(0,
- A ^t ^ A,
is a solution of (1).
To complete the definition of <p(f) for all values of t, i.e.
oo < t < oo, we could repeat the process of reflection, using
next the lines / = A and t = A;
anc* so on ' ^ ut ** *s s^ m P^er to
!
/i\ introduce the idea of periodicity.
\
v Periodicity. Let the function
now be extended to all values
-2A -A
of t by the requirement of peri-
FIG. 156 .
odicity
Hence
/i o\ '/ t\ ft\
(13) UQ = ^>(0,
namely
If w = ,
there is only one value, namely, t = 0; and if
w = &, then t A y
A. But only one should be counted,
A DIFFERENTIAL EQUATION 459
*i +7= tQ, 7i
= <o
- 'i-
(14) u *( + 7i)
is the solution we set out to obtain.
But is this solution unique, or are there still other solutions
which satisfy the same initial conditions? If a < u^ < 6, the
answer is affirmative for values of t near ^ but for remote values, ;
u = a, u = 6
(15)
(16)
= (u
- a) (6
- u)*t(u),
(J~J
ismore easily dealt with. A particular integral of (16) is given
by the formula
dM
= === a ^ u < b
f-(b _
(17) t ,
a
The inverse function,
(19) <p(- t)
= <p(t).
u= b =
(20) u ,(<),
(21) u = ?( + 7).
A further case, namely :
(22)
(23)
a)
- =+ (u
- a) (6
- t)vV(tt);
b)
f =-(u-a)(b-u)Vt(d.
t
* =
j<,
(25)
462 APPENDIX B
can be reduced by a linear transformation :
- -
U= 2u b-a
,
a 6
'
(26)
* 2
Equation (26) becomes, on suppressing the factor sin :
(28)
(29)
(30)
=
9
de -
(3D oo <e < oo.
"/: ^(cosfl)'
Write
(32)
where
d9
(33) f .
- 24.
_J V^(cose)
Then g(6) is periodic with the period 2v. For,
(34)
=
(36) t
^+
defines 6 as a single- valued function of t, since the integral (31)
represents a monotonic function of 0. Let be written in the
form:
(37) 0= J + /KO-
Then h(t) has the period 2 A :
For, let have an arbitrary value in (36) and let the correspond-
ing value of t be t :
,,*('. + 20 + , ( ,. + ar) .
By virtue of (35),
Hence
(40)
c =
Jf(x)dx.
Then
J + AL
f(x) dx
= C,
/V
Let
X
f(x)dx
-
jx.
c
<p(x + A) = <p(x).
For,
x+A
v(x + A) - *(x) =
ff(x)dx
-( x + A)+
A DIFFERENTIAL EQUATION 465
Hence
where
and
A
X =
C=ff(x)dx, |-
Xz +7 or X(z
- x n ),
A
= 0.
ff(x)dx
o
APPENDIX C
CHARACTERISTICS OF JACOBI'S EQUATION
Although Jacobi's partial differential equation of the first order :
AN
A)
d(Jr _ dH dp r _ m r-l,..-,m,
~dt~Wr ~dt~~Wr'
where // = H(q ly
-
, qm , ply , pm , t), we have not found it
L
T l'
.. ' n> >
.
'
dx
Let F(x ly - -
,
x n z, 2/i,
, 2/n), together with its partial deriva-
,
tives of the first two orders, be continuous for those values of the
arguments for which (x l9 x n z) is an interior point of an
, ,
*
(n + l)-dimensional region R
of the space of the variables
(x 19 ,
xn , z), and the yk are wholly unrestricted. Use the
notation :
/i\ y dF 7 dF v 3F
__
Z __
(1) Xk
~^' -~te>
Yk
~^
At a given point A :
(a,, ,
an , c, b 19 ,
b n) = (a, c, b) of R
let the Yk not all vanish ;
in particular, let Yn ^ 0.
"""
dxk dz dyk
II = = k =l
Yk S y k Yk Xk + yk Z* ' '
T.
^t
= f-(r
^i \^n , -^i
-r
,
.
,
r
-^n 1)
7
* 9
11
Ml 9 9
11
if* /9
^
i = l,---,n-l;
(2)
(4) F(x, - -
, 4-i, On, 2, 2/i, , )
= 0,
(5) z = *(*!,
-
-,*),
468 APPENDIX C
of the partial differential equation I. will lie. In the (hyper-)
plane x n = a n of this space let a manifold be defined by the
equation :
(6) 2 = Cdfo ,
' '
, Xn-l),
and
Furthermore, let
(8) (
1 z=F, xn = *.
As regards y n we ,
see from I. and A) that it is given by the equa-
tion:
(9) F ss H(x lt ,
xm * , , Vi,
- -
, 2/m, xn ) + yn = 0.
- dp r dy n
=
(10)
m i m d# dff
dq r dt
Xr /7
Q r
</r >
3
?/
Jf r
= 7iPr ,
r
r =s 1
1,
.
,
*n
771 ,
dq r
(12) ' ' m.
dt dp r dt dq r
(
\
.
/
dtf^m.
fit QJ J
at ot
dV =
(14)
dt Z PrQr -
Observe in passing that the right-hand side of (14) is the La-
grangean Function, L :
H+L =
5) pr qr ,
and so
(15)
~--LL
dV
-
and (14), we have not brought out the direct testimony of the
general theory in the present case. Let us turn back, then, to
Equations (2) and Condition (4) or (4'). It appears that the
solution of Equations (10) takes the form :
n __ / // . -<)
Qr Jr(J> , Pi
7)
Pr ~ Jn+r
/ (t
\f , >
n
Q.m ,
V ' 0>
(16)
, <7 m ,
F 0) Pm ,
470 APPENDIX C
where r = 1, ,
m= n 1 and
(17) y =- H (<?,, , qn , Pl , -, pm ,
t )
(18)
V ' '
(19) 2/n
=- //(ft,
-
, g, p,, , ?m, 0-
(30)
or
t
V =
to
(21) +
we choose w subject to the conditions under (6) :
(22) F = cofoo, , qm )
and set
n = r 's
3. Application. We
have seen in Chap. XV, 2, that Hamil-
ton's Equations can be solved by a contact transformation :
= d$ n d$
/f\n\
(23) Pr
w , Pr=~ W ,
r = l,--,m,
-i
Qr = ar , Pr = r, r = 1, , m,
CHARACTERISTICS OF JACOBFS EQUATION 471
dQ r _ Q d^-n r-1 m
r-l,---,m,
"rfT" '
~3T~'
are concerned.
The demands that the function Sfa, -
, qm , Q if Qm , ,
for the ar :
(25) oj
= aj ,
- - -
,
am = am ,
where a r = #r ,
6r = pr are an arbitrary set of initial values
of </r, p r.
dS dS d*S
' '
dq r dctr dq r doL 8
!, ,, , , m
(27) cofo ,
-
, qm )
= a^i + + <* m qm .
(1) H = H(q l9
- -
,<?m , p lf -
,pm ).
Jacobi's Equation now takes the form :
A/ ,
A')
_+ *
(3) V =- ht + W,
where
w= w( qi ..-,?) ,
r,x v( dW 8W\ ,
C) ff^,...,^-.,... ,__)-*.
The derivatives of H(q l}
-
, qm , p l9 , pm ) with respect to
the p r are not all 0.* Let
*
Then the equation
(6) Pi
= x(7i, ",?, A, P2i
'
Pm),
^
c)
dW = . dW dW
Let
Hence
pr = ar ,
r = 2, , m,
Since
r ? s
(J1S\/ = {
0,
r, s
= m,
\dardqa I 1, r = s
2, ,
Now,
(12) x i
98 8S
(14)
or
dW
(15)
= Pr, r = 2,
- - -
,
m ;
(16)
dq r dW -dp r
m.
(17) r-1,
^L 2 Pr
Sp r 8p r dq r
Since
m_~
dP r
.
9"
we have
dW
(18)
or
(19) W
(20) W = 2 Hdt + W Q.
CHARACTERISTICS OF JACOBFS EQUATION 475
equation :
fa 2Z\ n
,...,x n ,
2, ,..-, )=o.
Let
^( 2, ,a n ) = c,
^*(oa, , o) = 6*, A; = 2, ,
M.
Z = ^(Xj,
- -
,
X n ),
^(i, , On)
= c,
^/( a i>
' ' '
> n)
==
6/, y = 1, , n,
PATHS
Consider* a system of n particles m (x, y^ Zi) acted on by t :
A) niiXi Xi niiiji
= Yi mi'Zi = Z,-
determines a curve :
Xi = xt(t), yi
= 2/ t (0, *< = *;(0
(2) _
Xi ~ .
Vi
~
__
'
.
Zi "
__ ,
are arbitrarily assigned, not more than one path (2) exists which
(5) Xi = Xi (0, Vi
= Vi (0, Zi
= ^ (0,
all belong to R. Hence the curve (5) lies wholly in R.
478 APPENDIX D
Consider an arbitrary line through P, but not perpendicular
to the axis of
t. Let its direction components be a,-, ft, 7,-, K,
where K ^ 0. There may be a path corresponding to P, such
that at this point
Xi :
yi : Zi = <*< :
ft :
7*.
When this is not the case, not all lines through P correspond
to paths, and so certain relations between the direction com-
ponents (ca, pi, 7,, K) must exist. Thus we are led to a second
postulate.
+B +C 7 +D K = s =
(6) 2} (A 3i ai 8i ft ai t) 8 0, 1, , cr,
ii
where A 8 i, B 8i , C,, D 8 are functions* of (Xi, yl ,
z t , t) such that the
rank of the matrix :
(7)
is <r.
~~ ~~ ~~
i ft 7i ;
K
B) ^ (Ati Xi +B 8i yi + C, Zi) +D = 8 0, s = 1,
-
,
<r.
*
Throughout the whole treatment, the continuity of the functions which enter,
and the existence and continuity of such derivatives as it may be convenient to use,
are assumed.
GENERAL PROBLEM OF RATIONAL MECHANICS 479
(8)
(9)
= 0,
dt
C)
= /x + I.
II
D'ALEMBERT'S PRINCIPLE
In practice the equations which the Si, SK satisfy are ,
(10) 'Xi
= xi + A7, Yi = y; + 17; Zi = z; + z;,
where X\, FJ, Z\ are known in terms of the coordinates a:,-, y^ z,-,
Xi,ilij Zi,
t f an arbitrary point of D, ami where
i=l
i=l
Here, A'ai ,
B'a i, C^i are known functions of the above #,-, 7/ t-, z^
Xi 9 tit* *i) t> and the rank of the matrix :
(13)
(14) = 0,
GENERAL PROBLEM OF RATIONAL MECHANICS 481
where T/,-, f i
are 3n arbitrary quantities.
,-,
Under the sanction
of Postulate III. this equation can be replaced by the following :
~ + - Ffl* + (w2 - -
(15) 2) (m t i XI) fc (m<fr Z')f 0,
<-l
where ,-, ?7 t , f ,-
are any 3n quantities which satisfy the condition
(12).
Multiply the 0-th equation (12) by X^ and subtract the re-
- - - -
(16)
<=1
2) (mtfi X,' 5) A'fiiljh
0-1
+ (mtfr Y'<
% B'^n
/3-1
But the latter are arbitrary. Hence each coefficient must vanish.
The 3n v equations thus obtained express the result of elim-
inating the unknown S SK i.e. the X*, Yf, Zf, from the
l , ,
+ #* = = -
%(Eyi Xi + Fyi y< + 7 3n
.
E) Gyi 2,) 0, 1, , v,
<-i
Q Q
t
(xfj yfj Zi Xi yfj zP, Q ) is
,
an arbitrary point
,
of Z), Equations E)
unique.
Retrospect. These Postulates complete the formulation of the
class of problems in Rational Mechanics which we set out to
isolate. The role which d'Alembert's Principle * plays is two-
fold. First, it requires that the relations between the unknown
&* shall b e linear. Secondly, it performs the elimination
' m
S\, 9
tions (12). Hence the virtual displacements to,-, 6r/, fai of (17)
will not coincide save as to infinitesimals of higher order with any
in B) and the Apt, B'ftiy C'pi in (12) are respectively equal to each
* Historically d'Alembcrt's
Principle took its start in the assumption of a con-
dition, necessary and sufficient, that a system of forces, acting on a system of
particles, be in equilibrium, namely, that the virtual work corresponding to a
virtual velocity be nil. When a system of forces not in equilibrium acts on a system
of particles, the former can be replaced by a system of forces in equilibrium
through the introduction of "counter effective forces" or "forces of inertia"
(sic), and thus d'Alembert arrived at the General Equation of Dynamics.
GENERAL PROBLEM OF RATIONAL MECHANICS 483
other. But even so, if the D a are not all 0, the virtual displace-
ment will not tally save as to infinitesimals of higher order with
any possible actual displacement.
Finally it can happen that, in addition, the D, are all 0.
Then the virtual displacement corresponds to a possible displace-
ment. But this is a very special, though highly important, case.
Ill
LAGRANGE'S EQUATIONS
Let
', 0m,
(18) -, ffm,
-,ff,
dq,
(19)
(20) qr
= qr (t), r = 1, ...,m.
For the path in question we have :
dtdqr
__ Sq r
~ Q" '
where
<+ Yi+ 7
8xi
4- 8yi 4- 8z
(22)
r = 1, ,
m. For then
vhere the 8qr are arbitrary. Equations C), if present, are all
satisfied identically when the Xi, y^ Zi are expressed in terms of
"
tte q r and t by Equation (18). Equations B') are not present in the
problem. The system is, to be sure, holonomic, but it is not the
only case in which this is so.
are flee from the /S's,and that the S's coincide with the
X*, Yf) Z*. We now divide the S's into two categories :
(25) Q = r Qr + Q?, r = 1,
-
-, m,
where Q'r is known in terms of such values of qr qrj ,
t as correspond
to points of D.
In the particular case before us, namely, Equations (18), it
can happen that the equation :
(27) Oft! T t + - - -
+ a'tmKm = 0,
= 1, ,
v l9
(28)
is v l
'
9
and conversely, when Equations (27) hold, then (26) is true.
Equations B'), C) go over in the present case into :
== Oi == 1,
" " "
Bq) al*ll 4" ~T~ ttam^wt ~f" &a 0, ', /Zj 9
(29)
is /z t ;
the rank of the matrix :
(30)
Fg~
being ^.
486 APPENDIX D
(24) are too restricted. The &, rj if f. of Equations (11) are quan-
tities which must be able to take on every set of values which
satisfy (12). The {* which here figure, given by (24), are
.-, 77,-,
not free under the condition (24), but arc unwarrantably restricted
by (24).
In a given problem the desideratum usually is, to choose m as
small as possible, subject to the requirement that the same degree
of elimination of the'S's through (24) shall have been attained,
as if Equations (11) and (12) had been used.
IV
NOTES
Consider the dynamical system that consists of a bead sliding
on a fixed circular wire and acted on by no other forces than the
reaction of the wire. Equations A) take the form :
mx = X, my = Y, mz = Z.
Let the wire be a circle whose axis is the axis of 2. Then Equa-
tions B) become :
xx + yy =
B)
o
This system of Pfaffians is completely integrable :
f x 2
+ 2/
2 = a 2
I z = c
GENERAL PROBLEM OF RATIONAL MECHANICS 487
X = X*, Y = Y*,
and
X*+ F% =
provided
^ + y t] = 0.
Then
= X* (- a sin ?) + 7* (a cos ?)
= ;T(-2/) + F** = 0.
Hence, finally :
~ * Q'
Jt ~dij ~dq
is still true. But
488 APPENDIX D
We
have not enough equations to solve the problem. This is
the case in which Lagrange's Equations are said to "fail" or be
"inapplicable." The failure lies, not in Lagrange's Equations,
but in a misuse of them. We should take = 2. Let us first m
treat the problem, however, by the methods of Parts I., II., before
Lagrange's Equations were introduced in Part III. Here, then,
r72rr
m at* = X* = - R cos B + R sin
m-j%
= F* =-juftcos0 - Rsin0.
ut
x** + 7*77 = 0,
or
( R cos + p,R sin 6) + (- nR cos 6 - R sin 0) rj
= 0,
or, finally,
( x + y) + ( MZ - y) v
= 0,
/
(i*x
d x\
2
x* + y* = a2 ,
(31)
7<S
C fl ) r = a.
here becomes :
0,
*"l + M^2 = 0-
TTi
= /i, ^2
-. ma _
dO*
7ri +
,
ma _
d*0
or
#8 <W
M
Mass, Absolute unit of, 79 Operator, Symbolic vector, 254
Moments about centre of, 139, 205 Orbit of a planet, 111, 113, 435
Motion of the centre of, 120, 201 Oscillations, Small, 333
Notion of, 118 Osculating plane, 90, 92
494 INDEX
Rotation of the axes, 454
Parabolic motion, 93 about a fixed axis, 127, 136
Parallel forces in a plane, 21, 23 of a plane lamina, 139
in space, 36 of a rigid body, Chap. VI